id
stringlengths 20
52
| title
stringlengths 3
459
| abstract
stringlengths 0
12.3k
| classification_labels
list | numerical_classification_labels
list |
|---|---|---|---|---|
SCOPUS_ID:84888407059
|
A Fused Feature Extraction Approach to OCR: MLP vs. RBF
|
This paper is focused on evaluating the capability of MLP and RBF neural network classifier algorithms for performing handwritten character recognition task. Projection profile features for the character images are extracted and merged with the binarization features obtained after preprocessing every character image. The fused features thus obtained are used to train both the classifiers i.e. MLP and RBF Neural Networks. Simulation studies are examined extensively and the proposed fused features are found to deliver better recognition accuracy when used with RBF Network as a classifier.
|
[
"Visual Data in NLP",
"Information Extraction & Text Mining",
"Text Classification",
"Information Retrieval",
"Multimodality"
] |
[
20,
3,
36,
24,
74
] |
SCOPUS_ID:85079743706
|
A Fusion Model-Based Label Embedding and Self-Interaction Attention for Text Classification
|
Text classification is a pivotal task in NLP (Natural Language Processing), which has received widespread attention recently. Most of the existing methods leverage the power of deep learning to improve the performance of models. However, these models ignore the interaction information between all the sentences in a text when generating the current text representation, which results in a partial semantics loss. Labels play a central role in text classification. And the attention learned from text-label in the joint space of labels and words is not leveraged, leaving enough room for further improvement. In this paper, we propose a text classification method based on Self-Interaction attention mechanism and label embedding. Firstly, our method introduce BERT (Bidirectional Encoder Representation from Transformers) to extract text features. Then Self-Interaction attention mechanism is employed to obtain text representations containing more comprehensive semantics. Moreover, we focus on the embedding of labels and words in the joint space to achieve the dual-label embedding, which further leverages the attention learned from text-label. Finally, the texts are classified by the classifier according to the weighted labels representations. The experimental results show that our method outperforms other state-of-the-art methods in terms of classification accuracy.
|
[
"Semantic Text Processing",
"Text Classification",
"Representation Learning",
"Information Retrieval",
"Information Extraction & Text Mining"
] |
[
72,
36,
12,
24,
3
] |
SCOPUS_ID:85065924691
|
A Fuzzy Approach for Sentences Relevance Assessment in Multi-document Summarization
|
Text summarization is becoming an indispensable solution for dealing with the exponential growth of textual and unstructured information in digital format. In this paper, an unsupervised method for extractive multi-document summarization is presented. This method combines the use of a semantic graph for representing textual contents and identify the most relevant topics with the processing of several sentences features applying a fuzzy logic perspective. A fuzzy aggregation operator is applied in the sentences relevance assessment process as a contribution to the multi-document summarization process. The method was evaluated with the Spanish and English texts collection of MultiLing 2015. The obtained results were measured through ROUGE metrics and compared with those obtained by other solutions reported from MultiLing2015.
|
[
"Summarization",
"Text Generation",
"Information Extraction & Text Mining"
] |
[
30,
47,
3
] |
SCOPUS_ID:0141867824
|
A Fuzzy Approach to Classification of Text Documents
|
This paper discusses the classification problems of text documents. Based on the concept of the proximity degree, the set of words is partitioned into some equivalence classes. Particularly, the concepts of the semantic field and association degree are given in this paper. Based on the above concepts, this paper presents a fuzzy classification approach for document categorization. Furthermore, applying the concept of the entropy of information, the approaches to select key words from the set of words covering the classification of documents and to construct the hierarchical structure of key words are obtained.
|
[
"Information Retrieval",
"Text Classification",
"Information Extraction & Text Mining"
] |
[
24,
36,
3
] |
https://aclanthology.org//1995.iwpt-1.5/
|
A Fuzzy Approach to Erroneous Inputs in Context-Free Language Recognition
|
Using fuzzy context-free grammars one can easily describe a finite number of ways to derive incorrect strings together with their degree of correctness. However, in general there is an infinite number of ways to perform a certain task wrongly. In this paper we introduce a generalization of fuzzy context-free grammars, the so-called fuzzy context-free K-grammars, to model the situation of malting a finite choice out of an infinity of possible grammatical errors during each context-free derivation step. Under minor assumptions on the parameter K this model happens to be a very general framework to describe correctly as well as erroneously derived sentences by a single generating mechanism. Our first result characterizes the generating capacity of these fuzzy context-free K-grammars. As consequences we obtain: (i) bounds on modeling grammatical errors within the framework of fuzzy context-free grammars, and (ii) the fact that the family of languages generated by fuzzy context-free K-grammars shares closure properties very similar to those of the family of ordinary context-free languages. The second part of the paper is devoted to a few algorithms to recognize fuzzy context-free languages: viz. a variant of a functional version of Cocke-Younger-Kasami’s algorithm and some recursive descent algorithms. These algorithms tum out to be robust in some very elementary sense and they can easily be extended to corresponding parsing algorithms.
|
[
"Syntactic Parsing",
"Syntactic Text Processing"
] |
[
28,
15
] |
SCOPUS_ID:85114692220
|
A Fuzzy Approach to Language Universals for NLP
|
One of the currently biggest challenges in NLP is to develop multilingual language technology. Lack of data in low-resources languages poses great difficulty to NLP researchers and limits NLP technology's availability to a small number of resource-rich languages. It has been shown that linguistic typology and the knowledge of language universals can help NLP in the development of multilingual resources. To contribute to this research area, we present a fuzzy approach to language universals. Our proposal combines a constraint-based formalism with fuzzy logic to define a fuzzy-gradient model to characterize linguistic universals. This model will allow us to evaluate linguistic universals and to define a universal grammar. This universal grammar will be integrated into an automatic technique to infer from linguistic data the particular grammar of any understudied natural language.
|
[
"Typology",
"Syntactic Text Processing",
"Multilinguality"
] |
[
45,
15,
0
] |
SCOPUS_ID:85063949881
|
A Fuzzy Approach to Text Classification with Two-Stage Training for Ambiguous Instances
|
Sentiment analysis is a very popular application area of text mining and machine learning. The popular methods include support vector machine, naive bayes, decision trees, and deep neural networks. However, these methods generally belong to discriminative learning, which aims to distinguish one class from others with a clear-cut outcome, under the presence of ground truth. In the context of text classification, instances are naturally fuzzy (can be multilabeled in some application areas) and thus are not considered clear-cut, especially given the fact that labels assigned to sentiment in text represent an agreed level of subjective opinion for multiple human annotators rather than indisputable ground truth. This has motivated researchers to develop fuzzy methods, which typically train classifiers through generative learning, i.e., a fuzzy classifier is used to measure the degree to which an instance belongs to each class. Traditional fuzzy methods typically involve generation of a single fuzzy classifier and employ a fixed rule of defuzzification outputting the class with the maximum membership degree. The use of a single fuzzy classifier with the above-fixed rule of defuzzification is likely to get the classifier encountering the text ambiguity situation on sentiment data, i.e., an instance may obtain equal membership degrees to both the positive and negative classes. In this paper, we focus on cyberhate classification, since the spread of hate speech via social media can have disruptive impacts on social cohesion and lead to regional and community tensions. Automatic detection of cyberhate has thus become a priority research area. In particular, we propose a modified fuzzy approach with two-stage training for dealing with text ambiguity and classifying four types of hate speech, namely, religion, race, disability, and sexual orientation - and compare its performance with those popular methods as well as some existing fuzzy approaches, while the features are prepared through the bag-of-words and word embedding feature extraction methods alongside the correlation-based feature subset selection method. The experimental results show that the proposed fuzzy method outperforms the other methods in most cases.
|
[
"Information Extraction & Text Mining",
"Text Classification",
"Ethical NLP",
"Sentiment Analysis",
"Information Retrieval",
"Responsible & Trustworthy NLP"
] |
[
3,
36,
17,
78,
24,
4
] |
SCOPUS_ID:85055867662
|
A Fuzzy Decision Support Model with Sentiment Analysis for Items Comparison in e-Commerce: The Case Study of http://PConline.com
|
Decision support is a vital function in electronic commerce (e-commerce). The purpose of this paper is to construct a review-based decision support model for items comparison in e-commerce. The proposed model uses probability multivalued neutrosophic linguistic numbers (PMVNLNs) to characterize online reviews. It overcomes the limitation of existing models by considering neutral information and hesitancy in text reviews. The fuzzy characterization of reviews (i.e., PMVNLN) can reflect similarities and differences in positive (negative) information. In addition, the model considers consumers' bounded rational behaviors by combining the regret theory with an outranking method. We empirically compare the proposed model with models in http://PConline.com and four existing models with data from http://PConline.com. The performance of these models in terms of accuracy is measured by the total relative difference metric. Results indicate the good performance of the proposed model. Our model is a promising option for e-commerce to provide consumers with good decision support service.
|
[
"Sentiment Analysis"
] |
[
78
] |
SCOPUS_ID:85140345026
|
A Fuzzy Declarative Approach to Classify Unlabeled Short Texts Based on Automatically Constructed WordNet Ontologies
|
In this paper, we present an approach to categorizing short texts which only require, as user input, the category names defined using an ontology of terms automatically extracted from WordNet and modeled as a set of proximity equation. The use of a fuzzy extension of Prolog allows us to develop an efficient text classifier in which the classification method is explainable. The results of the experiments showed that the proposed method achieved a reasonable good performance.
|
[
"Low-Resource NLP",
"Semantic Text Processing",
"Information Retrieval",
"Information Extraction & Text Mining",
"Knowledge Representation",
"Text Classification",
"Responsible & Trustworthy NLP"
] |
[
80,
72,
24,
3,
18,
36,
4
] |
SCOPUS_ID:85115247071
|
A Fuzzy Deep Learning Approach to Health-Related Text Classification
|
Following the tremendous amounts of text generated in social networks and news channels, and gaining valuable and dependable insights from diverse sources of information is a tedious task. The challenge is increased during specific periods, for example, in a pandemic event like Covid-19. Existing text categorization methods, such as sentiment classification, aim to help people tackle this challenge by categorizing and summarizing the text content. However, the inherent uncertainty of user-generated text limits their efficiency. This paper proposes a novel architecture based on fuzzy inference and deep learning for sentiment classification that overcomes this limitation. We evaluate the proposed method by applying it to well-known health-related text datasets and comparing the accuracy with state-of-the-art methods. The results show that the proposed fuzzy fusion methods increase the accuracy compared to individual pretrained models. The model also provides an expressive architecture for health news classification.
|
[
"Information Extraction & Text Mining",
"Information Retrieval",
"Text Classification",
"Sentiment Analysis"
] |
[
3,
24,
36,
78
] |
SCOPUS_ID:85136800195
|
A Fuzzy Grammar for Evaluating Universality and Complexity in Natural Language
|
The paper focuses on linguistic complexity and language universals, which are two important and controversial issues in language research. A Fuzzy Property Grammar for determining the degree of universality and complexity of a natural language is introduced. In this task, the Fuzzy Property Grammar operated only with syntactic constraints. Fuzzy Natural Logic sets the fundamentals to express the notions of universality and complexity as evaluative expressions. The Fuzzy Property Grammar computes the constraints in terms of weights of universality and calculates relative complexity. We present a proof-of-concept in which we have generated a grammar with 42B syntactic constraints. The model classifies constraints in terms of low, medium, and high universality and complexity. Degrees of relative complexity in terms of similarity from a correlation matrix have been obtained. The results show that the architecture of a Universal Fuzzy Property Grammar is flexible, reusable, and re-trainable, and it can easily take into account new sets of languages, perfecting the degree of universality and complexity of the linguistic constraints as well as the degree of complexity between languages.
|
[
"Text Complexity",
"Semantic Text Processing",
"Syntactic Text Processing"
] |
[
42,
72,
15
] |
SCOPUS_ID:85043337802
|
A Fuzzy Logic Approach to Predict the Popularity of a Presidential Candidate
|
We are noticing a new era of social networks where in a blink of eye millions of tweets about any topic can be emerged. Especially, when an event like national election comes for a nation, the messages in social media especially twitter rises at its peak. The amount of data twitter has during that time is enormous and those tweets were never been used to analyze anyone’s popularity. Our work is focused on predicting a presidential candidate’s live popularity through sentiment analysis. We design the system to predict the popularity by a single day. To do this several features from tweets of a particular day have been passed through a dimensionality reduction algorithm, e.g., PCA (Principal Component Analysis). Consequently, the PCA components have been exercised into a fuzzy system. In particular, we used ANFIS (Adaptive Neuro Fuzzy Inference System) to predict a presidential candidate’s popularity on a single day.
|
[
"Sentiment Analysis"
] |
[
78
] |
SCOPUS_ID:85064043471
|
A Fuzzy Logic Model for Evaluating Customer Loyalty in e-Commerce
|
This research proposes a model for customer’s loyalty by sentiment analysis of ecommerce products. The purpose behind this research is to evaluate the response of customers in the shortest time. It practices sentiment analysis that tends to understand the user’s feedback about the product and services on ecommerce sites. The data is openly available on these sites in the form of reviews, comments and appraisals. This data focus on the customer’s opinions and helps business to take proficient decisions in the limited time. It takes subjective reviews because objective part contains emotion symbols. Many people do not know the proper use of emotions. It also prefers to use Stanford POS (Parts-of-Speech) tagger from Stanford Core NLP toolkit. This tagger assigns part of speech to every word of the reviews as we extract adjectives to measure the scores. This paper also used these techniques: tokenization, Lemmatization and stop words removal. By the use of soft computing approach- Fuzzy logic, it will able to design a customer loyalty model by its membership functions and truth values between 0 and 1. It uses SentiWordNet software to measure the P-N polarity scores. This proposed model reduces the problems from the related past researches. This research collects results from the reviews from Amazon.com which shows 72% customers are loyal towards ecommerce products. The outcomes that can allow business organization improve customer loyalty techniques to gain profitable results.
|
[
"Polarity Analysis",
"Sentiment Analysis"
] |
[
33,
78
] |
SCOPUS_ID:85111250447
|
A Fuzzy Near Neighbors Approach for Arabic Text Categorization Based on Web Mining Technique
|
Nowadays, the quantity of textual content available online has experienced such a colossal increase. Hence, the need for a system to investigate this content data is mandatory. In this concern, Text Categorization (TC) highlights many performance methods and techniques to analyze, explore and classify various types of documents. This study consists of two main steps. First, we extract terms from text documents using Fuzzy Near Neighbors (FNN) with web-based mining techniques algorithm. Second, we identify documents according to a particular form of similarity based on combining all Arabic encyclopedic dictionaries using clustering algorithms. In this article, Fuzzy C-Means (FCM) as a clustering algorithm is used to perform the precision of documents’ classification. This work suggests Arabic TC based on a multilingual encyclopedic dictionary (Arabic WordNet, OMW, Wikipedia, OmegaWiki, Wictionary, and Wikidata). To evaluate the efficacy of TC approach with FNN and FCM, an experimental study using a real-world dataset is carried out. The results of the present study indicate that proposed approach outperforms the traditional one and produces good results.
|
[
"Information Extraction & Text Mining",
"Information Retrieval",
"Text Classification",
"Text Clustering"
] |
[
3,
24,
36,
29
] |
SCOPUS_ID:84902668800
|
A Fuzzy PROMETHEE Approach for Mining Customer Reviews in Chinese
|
Online customer reviews of products have a great impact on potential customers' purchase decisions and provide valuable customer opinions to businesses. However, it is difficult for a customer to go through the huge number of customer reviews of a product to make an informed decision. The opinion comparison, one of the important tasks in opinion mining, uses main product features that have been commented upon by consumers to compare competing products. Because the task of comparing customer opinions can be expressed as the ranking of alternative products using key product features, it can be modeled as a multi-criteria decision making (MCDM) problem. The goal of this paper is to propose fuzzy PROMETHEE, an MCDM method, to rank alternative products based on online customer reviews of products. An experiment is designed to test the proposed method using a sample of Chinese reviews of mobile phones. The results demonstrate that this approach can not only generate a reliable and realistic ranking of products, but also identify key product features that are considered by consumers as the most important aspects of a product. © 2014 King Fahd University of Petroleum and Minerals.
|
[
"Opinion Mining",
"Sentiment Analysis"
] |
[
49,
78
] |
SCOPUS_ID:0028721477
|
A Fuzzy Reasoning Database Question Answering System
|
The present paper describes a question answering system based on fuzzy logic. The proposed system provides the capability to assess whether a database contains information pertinent to a subject of interest by evaluating each comment in the database via a fuzzy evaluator that attributes a fuzzy membership value indicating its relationship to such subject. An assessment is provided for the database as a whole regarding its pertinence to the subject of interest, and consequently comments that are considered irrelevant to the subject may be discarded. The system has been developed for the examination of databases that were created during the development of the IBM 4381 computer systems, for bookkeeping purposes, to assess whether such databases contain information pertinent to the functional changes that occurred during the development cycle. The system, however, can be applied with minimal changes to a variety of circumstances, provided that the fundamental assumptions for the development of the membership functions are respected in the new application. Its applicability, without modifications, assuming the same subject of interest, is granted for databases comprising similar characteristics to that of the original database for which the system has been developed. © 1994 IEEE
|
[
"Natural Language Interfaces",
"Reasoning",
"Question Answering"
] |
[
11,
8,
27
] |
SCOPUS_ID:85070710312
|
A Fuzzy Reasoning Process for Conversational Agents in Cognitive Cities
|
Facing the challenges in a city that is to be understood as a complex construct, this article presents a solution approach for the further development of existing conversational agents, which should be used particularly in cities, for instance, as a source of information. The proposed framework consists of a fuzzy analogical reasoning process (based on structure-mapping theory) and a network-like memory (i.e., fuzzy cognitive maps stored in graph databases) as additions to the general architecture of a chatbot. Thus, it represents a concept of a global fuzzy reasoning process, which allows conversational agents to emulate human information processing by using cognitive computing (consisting of soft computing methods and cognition and learning theories). The framework is already in the third iteration of its development. Three experiments were conducted to examine the stability of the theoretical foundation as well as the potential of the framework.
|
[
"Natural Language Interfaces",
"Reasoning",
"Dialogue Systems & Conversational Agents"
] |
[
11,
8,
38
] |
SCOPUS_ID:85130684578
|
A Fuzzy System for Identifying Partial Reduplication
|
Reduplication is a common feature more or less in almost all languages. It is a linguistic process that has been studied since back in both a morphological as well as a phonological process. Literary reduplication is the repetition of tokens in various forms like morpheme, phrase, word, etc. In most cases, it affects the syntactic or/and semantic meaning of the original words. But in several cases, it is not the exact reduplication rather partial reduplication. In the exact or total reduplication, it exactly reiterates a word or a phrase (e.g. fifty-fifty, bye-bye in English) while in partial reduplication reiteration partially (e.g. flip-flop in English). The morphological structure of a total or partial reduplication is relatively simple for English-like languages, but it is complex in the case of African, Austronesian, or South-Asian languages. Earlier researchers tried to address this problem using the various heuristic approaches. This paper presented a set of fuzzy-based approach to deal with partial reduplication in the Bengali language.
|
[
"Syntactic Text Processing",
"Morphology"
] |
[
15,
73
] |
SCOPUS_ID:85137562965
|
A Fuzzy Training Framework for Controllable Sequence-to-Sequence Generation
|
The generation of music lyrics by artificial intelligence (AI) is frequently modeled as a language-targeted sequence-to-sequence generation task. Formally, if we convert the melody into a word sequence, we can consider the lyrics generation task to be a machine translation task. Traditional machine translation tasks involve translating between cross-lingual word sequences, whereas music lyrics generation tasks involve translating between music and natural language word sequences. The theme or key words of the generated lyrics are usually limited to meet the needs of the users when they are generated. This requirement can be thought of as a restricted translation problem. In this paper, we propose a fuzzy training framework that allows a model to simultaneously support both unrestricted and restricted translation by adopting an additional auxiliary training process without constraining the decoding process. This maintains the benefits of restricted translation but greatly reduces the extra time overhead of constrained decoding, thus improving its practicality. The experimental results show that our framework is well suited to the Chinese lyrics generation and restricted machine translation tasks, and that it can also generate language sequence under the condition of given restricted words without training multiple models, thereby achieving the goal of green AI.
|
[
"Multilinguality",
"Language Models",
"Machine Translation",
"Semantic Text Processing",
"Speech & Audio in NLP",
"Text Generation",
"Cross-Lingual Transfer",
"Multimodality"
] |
[
0,
52,
51,
72,
70,
47,
19,
74
] |
SCOPUS_ID:85112674971
|
A Fuzzy Word Similarity Measure for Selecting Top-k Similar Words in Query Expansion
|
Top-$ k$ words selection is a technique used to detect and return the $ k$ most similar words to a given word from a candidate set. This is a crucial and widely used tool in various tasks. The key issue in top-$k$ words selection is how to measure the similarity between words. One popular and effective solution is to use a word embedding-based similarity measure, which represents words as low-dimensional vectors and measures the similarities between words according to the similarity of the vectors, using a metric. However, most word embedding methods only consider the local proximity properties of two words in a corpus. To mitigate this issue. In this article, we propose to use association rules for measuring word similarity at a global level, and a fuzzy similarity measure for top-k words selection that jointly encodes the local and the global similarities. Experiments on a real-world query task with three benchmark datasets, i.e., TREC-disk 4&5, WT10G, and RCV1, demonstrate the efficiency of the proposed method compared to several state-of-the-art baselines.
|
[
"Semantic Text Processing",
"Representation Learning"
] |
[
72,
12
] |
SCOPUS_ID:85084973685
|
A Fuzzy, Incremental and Semantic Trending Topic Detection in Social Feeds
|
Nowadays, a huge number of people participating in social networks is triggering a fast and wide spectrum of topics. Such trending topics are usually derived from the most frequent searches, the published posts and the daily news. The automated analysis for such data requires topics detection and tracking methods. Many challenges are being faced. It is difficult to discover the semantic relatedness when the same event is presented by different titles and to handle merging semantically identical topics from different channels (aggregation). Other hardships are the vagueness regarding the vast web collection, the scalability to analyze them, and the fact that it is a time consuming task. The framework introduced in this paper aims to solve these issues. Because a web document often consists of several topics, the suggested model employs a fuzzy C-Means (FCM) clustering based trending topics detection. It applies a semantic document similarity algorithm to resolve such ambiguity issues caused by the usage of synonyms, homonyms or different abstraction levels. This algorithm is also used to summarize the long documents. Furthermore, an incremental clustering technique is utilized to preserve high cohesiveness up-to-date top trending topics. The experimental results finally illustrate the effectiveness and the superiority of this model, compared with other trending topics detection algorithms, in terms of entropy and F-score measures.
|
[
"Information Extraction & Text Mining",
"Text Clustering"
] |
[
3,
29
] |
SCOPUS_ID:84857822750
|
A GA-based learning algorithm for inducing M-of-N-like text classifiers
|
This paper describes an extension of the classical M-of-N approach to text classification. The proposed hypothesis language is called M-of-N+. One distinguishing aspect of this language is its lattice-like structure, which defines a natural ordering in the hypothesis space useful to design effective search operators. To induce M-of-N+ concepts, a task-dependent Genetic Algorithm (called GAMoN), which exploits the structural properties of the hypothesis space, is proposed. In experiments on 6 standard, real-world text data sets, we compared GAMoN with one genetic rule induction method, namely, GAssist, and four classical non-evolutionary algorithms, notably, linear SVM, C4.5, Ripper and multinomial Naive Bayes. Experimental results demonstrate the effectiveness of the proposed approach. © 2011 IEEE.
|
[
"Information Retrieval",
"Text Classification",
"Information Extraction & Text Mining"
] |
[
24,
36,
3
] |
SCOPUS_ID:85144969107
|
A GAN-BERT Based Approach for Bengali Text Classification with a Few Labeled Examples
|
Basic machine learning algorithms or transfer learning models work well for language categorization, but these models require a vast volume of annotated data. We need a better model to tackle the problem because labeled data is scarce. This problem may have a solution in GAN-BERT. To classify Bengali text, we have developed a GAN-BERT based model, which is an adapted version of BERT. We have used two different datasets for this purpose. One is a hate speech dataset, while the other is a fake news dataset. To understand how the GAN-BERT and traditional BERT models behave with Bangla datasets, we have experimented with both. With a small quantity of data, we are able to get a satisfactory result using GAN-BERT. We have also demonstrated how the accuracy increases as the number of training samples increases. A comparison of performance between traditional BERT based Bangla-BERT and our GAN-Bangla-BERT model is also shown here, where we can see how these models react to a small number of labeled data.
|
[
"Language Models",
"Semantic Text Processing",
"Text Classification",
"Information Retrieval",
"Information Extraction & Text Mining"
] |
[
52,
72,
36,
24,
3
] |
SCOPUS_ID:85073067095
|
A GAN-based transfer learning approach for sentiment analysis
|
Transfer learning is an important artificial intelligence approach which extracts knowledge from source domain to solve tasks in the target domain. As a research hot topic, Generative Adversarial Networks (GAN) provides a powerful framework in constructing unsupervised models. A GAN consists of two neural networks: a discriminator to distinguish natural and generated samples, and a generator to deceive the discriminator. Generally, sentiment analysis of text is a big challenge in the Natural Language Processing (NLP). In this paper, A GAN-based transfer learning approach for sentiment analysis of cross-domain texts is presented. Experiments are performed in cross-domain e-commerce reviews. The results compiled demonstrate the effectiveness of the proposed approach.
|
[
"Language Models",
"Semantic Text Processing",
"Robustness in NLP",
"Sentiment Analysis",
"Responsible & Trustworthy NLP"
] |
[
52,
72,
58,
78,
4
] |
SCOPUS_ID:85135031214
|
A GAT-Based Chinese Text Classification Model: Using of Redical Guidance and Association Between Characters Across Sentences
|
Cognitive psychology research has shown that humans tend to use objects in the abstract real world as cognition. For the Chinese in particular, the first thing to emerge from the process of writing formation is pictographs. This feature is very useful for Chinese text classification. Fortunately, many basic and extended radical-related systems are also included in all Chinese dictionaries. Moreover, for Chinese texts the graph attention structure can better record the information of features in Chinese texts. To this end, we propose a GAT- based Chinese text classification model considering character, word, sentence position information, character position information and radicals as nodes in Chinese text. The model is based on Redical guidance and association between Characters across Sentences (RCS). In order to better exploit the common features of Chinese characters and sequence features in text, this paper designs a model with context-awareness, which is capable of collecting information at a distance. Finally, we conduct extensive experiments, and the experimental results not only prove the superiority of our model, but also verify the effectiveness of the radical and GAT model in Chinese text classification tasks.
|
[
"Information Retrieval",
"Text Classification",
"Information Extraction & Text Mining"
] |
[
24,
36,
3
] |
SCOPUS_ID:85134011575
|
A GAUSSIAN MIXTURE MODEL FOR DIALOGUE GENERATION WITH DYNAMIC PARAMETER SHARING STRATEGY
|
Existing dialog models are trained with data in an encoder-decoder framework with the same parameters, ignoring the multinomial distribution nature in the dataset. In fact, model improvement and development commonly requires fine-grained modeling on individual data subsets. However, collecting a labeled fine-grained dialogue dataset often requires expert-level domain knowledge and therefore is difficult to scale in the real world. As we focus on better modeling multinomial data for dialog generation, we study an approach that combines the unsupervised clustering and generative model together with a GMM (Gaussian Mixture Model) based encoder-decoder framework. Specifically, our model samples from the prior and recognition distributions over the latent variables by a Gaussian mixture network and the latent layer with the capability to form multiple clusters. We also introduce knowledge distillation to guide and improve the clustering results. Finally, we use a dynamic parameter sharing strategy conditioned on different labels to train different decoders. Experimental results on a widely used dialogue dataset verify the effectiveness of the proposed method.
|
[
"Language Models",
"Semantic Text Processing",
"Dialogue Response Generation",
"Natural Language Interfaces",
"Text Clustering",
"Text Generation",
"Dialogue Systems & Conversational Agents",
"Information Extraction & Text Mining"
] |
[
52,
72,
14,
11,
29,
47,
38,
3
] |
SCOPUS_ID:85145259278
|
A GDPR Compliant Approach to Assign Risk Levels to Privacy Policies
|
Data privacy laws require service providers to inform their customers on how user data is gathered, used, protected, and shared. The General Data ProtectionRegulation (GDPR) is a legal framework that provides guidelines for collecting and processing personal information from individuals. Service providers use privacy policies to outline the ways an organization captures, retains, analyzes, and shares customers' data with other parties. These policies are complex and written using legal jargon; therefore, users rarely read them before accepting them. There exist a number of approaches to automating the task of summarizing privacy policies and assigning risk levels. Most of the existing approaches are not GDPR compliant and use manual annotation/labeling of the privacy text to assign risk level, which is time-consuming and costly. We present a framework that helps users see not only data practice policy compliance with GDPR but also the risk levels to privacy associated with accepting that policy. The main contribution of our approach is eliminating the overhead cost of manual annotation by using the most frequent words in each category to create word-bags, which are used with Regular Expressions and Pointwise Mutual Information scores to assign risk levels that comply with the GDPR guidelines for data protection. We have also developed a web-based application to graphically display risk level reports for any given online privacy policy. Results show that our approach is not only consistent with GDPR but performs better than existing approaches by successfully assigning risk levels with 95.1% accuracy after assigning data practice categories with an accuracy rate of 79%.
|
[
"Ethical NLP",
"Responsible & Trustworthy NLP"
] |
[
17,
4
] |
SCOPUS_ID:85135555379
|
A GENERAL APPROACH FOR MEETING SUMMARIZATION: FROM SPEECH TO EXTRACTIVE SUMMARIZATION
|
Developing technologies and techniques have increased the amount of information and enabled easier access to information resources. However, due to the ever-growing amount of information sources, it has become difficult to access the information needed in a limited time. Consequently, the need for summary information has become important. This research is focused on the extraction of inferential written summaries of communications that occur in oral environments such as meetings, lectures and conferences. However, since this type of problem requires conversion from audio to text, it also includes issues such as the human factor, sound recording environments, and language-specific problems. This study aimed to take the audio recordings of the meetings, especially the IT sector, to process and summarize. Spontaneous conversations were converted into audio recordings and the obtained texts were summarized using extractive summarization techniques. The motivation of the study is to catch the important points that may escape the attention of the individuals at the meeting and to summarize the main agenda items for the personnel who could not attend the meeting. The experimentally generated dataset (converted from audio recordings to text) was summarized by three different human summarizers and compared with the summaries obtained from the developed inferential summative model. The results obtained are remarkable and it is seen that approximately 71% success was achieved.
|
[
"Speech & Audio in NLP",
"Summarization",
"Multimodality",
"Text Generation",
"Information Extraction & Text Mining"
] |
[
70,
30,
74,
47,
3
] |
SCOPUS_ID:85134029798
|
A GENERALIZED HIERARCHICAL NONNEGATIVE TENSOR DECOMPOSITION
|
Nonnegative matrix factorization (NMF) has found many applications including topic modeling and document analysis. Hierarchical NMF (HNMF) variants are able to learn topics at various levels of granularity and illustrate their hierarchical relationship. Recently, nonnegative tensor factorization (NTF) methods have been applied in a similar fashion in order to handle data sets with complex, multi-modal structure. Hierarchical NTF (HNTF) methods have been proposed, however these methods do not naturally generalize their matrix-based counterparts. Here, we propose a new HNTF model which directly generalizes a HNMF model special case, and provide a supervised extension. We also provide a multiplicative updates training method for this model. Our experimental results show that this model more naturally illuminates the topic hierarchy than previous HNMF and HNTF methods.
|
[
"Topic Modeling",
"Information Extraction & Text Mining"
] |
[
9,
3
] |
SCOPUS_ID:84923614552
|
A GENERIC FRONT END FOR TEXT-TO-SPEECH SYNTHESIS SYSTEMS
|
This paper describes how the SPRUCE (Speech Response from Unconstrained English) system is capable of acting as a "front end" to any text-to-speech system whether it be based on diphones, phonemes, demi-syllables, syllables or even words. This is possible because the architecture of SPRUCE includes a large dictionary which contains phonetic information of individual words along with syllable, stress and other prosodic information. This representation is such that the basic linguistic information for driving any text-to-speech synthesiser is already contained within the SPRUCE synthesis system. The advantage of using SPRUCE as a front end for a synthesiser is that it enables the synthesis system to make use of SPRUCE'S large word dictionary and its natural sounding intonation algorithm.
|
[
"Speech & Audio in NLP",
"Multimodality"
] |
[
70,
74
] |
SCOPUS_ID:85124613041
|
A GERMAN “LINGUISTIC ISLAND” OR A LINGUISTICALLY MIXED REGION? MULTILINGUAL PRACTICES IN THE KOČEVSKA (GOTTSCHEE) AREA
|
The article calls into question the understanding of the Kočevska (Gottschee) area as a “German language island”. Through examples of the use of different languages before the SecondWorldWar, it shows a different –multilingual or multicultural– image of this region. The author draws data from historical andarchival sources, as well as from a survey that she conductedamong Gottscheers (“Gottschee Germans”) living in Slovenia, Austria, Germany, the USAandCanada.
|
[
"Multilinguality"
] |
[
0
] |
SCOPUS_ID:84975721541
|
A GF miniature resource grammar for Tswana: modelling the proper verb
|
The Grammatical Framework (GF) not only offers state of the art grammar-based machine translation support between an increasing number of languages through its so-called Resource Grammar Library, but is also fast becoming a de facto framework for developing multilingual controlled natural languages (CNLs). For a natural language to share maximally in the opportunities that GF-based multilingual CNL support presents, it has to have a GF resource grammar. Tswana, an agglutinating Bantu language, spoken in Southern Africa as one of the eleven official languages of South Africa, does not yet have such a grammar. This article reports on the development of a so-called miniature resource grammar, a first step towards a full resource grammar for Tswana. The focus is on the modelling of the Tswana proper verb as it occurs in simple sentences. The (proper) verb is the morphologically most complex word category in Tswana, and therefore constitutes a notable contribution towards the development of a GF resource grammar for Tswana. The computational model is discussed in some detail, implemented and tested on a systematically constructed treebank.
|
[
"Multilinguality"
] |
[
0
] |
SCOPUS_ID:84915818700
|
A GIS anchored system for clustering discrete data points – A connected graph based approach
|
Clustering is considered as one of the most important unsupervised learning problem which groups a set of data objects, in such way, so that the data objects belongs to the same group (known as cluster) are very similar to each other, compared to the data objects in another group (i.e. clusters). There is a wide variety of real world application area of clustering. In data mining, it identifies groups of related records, serving as the basis for exploring more detailed relationships. In text mining it is heavily used for categorization of texts. In marketing management, it helps to group customers of similar behaviors. The technique of clustering is also heavily being used in GIS. In case of city-planning, it helps to identify the group of vacant lands or houses or other resources, based on their type, value, location etc. To identify dangerous zones based on earth-quake epi-centers, clustering helps a lot. In this paper, a set of data objects are clustered using two connected graph based techniques – MST based clustering and Tree Based clustering. After considering a lot of test cases, at the end of the paper, the second technique is found to be more suitable for clustering than the first one.
|
[
"Low-Resource NLP",
"Information Extraction & Text Mining",
"Structured Data in NLP",
"Text Clustering",
"Responsible & Trustworthy NLP",
"Multimodality"
] |
[
80,
3,
50,
29,
4,
74
] |
SCOPUS_ID:51449099268
|
A GIS-like training algorithm for log-linear models with hidden variables
|
Conditional random fields (CRFs) are often estimated using an entropy based criterion in combination with Generalized Iterative Scaling (GIS). GIS offers, upon others, the immediate advantages that it is locally convergent, completely parameter free, and guarantees an improvement of the criterion in each step. GIS, however, is limited in two aspects. GIS cannot be applied when the model incorporates hidden variables, and it can only be applied to optimize the Maxmimum Mutual Information Criterion (MMI). Here, we extend the GIS algorithm to resolve these two limitations. The new approach allows for training log-linear models with hidden variables and optimizes discriminative training criteria different from Maximum Mutual Information (MMI), including Minimum Phone Error (MPE). The proposed GIS-like method shares the above-mentioned theoretical properties of GIS. The framework is tested for optical character recognition on the USPS task, and for speech recognition on the Sietill task for continuous digit string recognition. ©2008 IEEE.
|
[
"Visual Data in NLP",
"Multimodality"
] |
[
20,
74
] |
SCOPUS_ID:77958071591
|
A GML compression approach based on on-line semantic clustering
|
Geography Markup Language (GML) has become a de facto international encoding standard for exchanging geospatial data among heterogeneous Geographic Information Systems (GIS). Whereas, structurally redundant tags and textual data representation usually inflate the sizes of GML documents substantially, which makes the storage and transport costly. In this paper, we propose an effective compression approach based on on-line semantic clustering of GML documents. The approach deals with a GML document under compression on the fly via separating data from structures, clustering data based on the semantic similarities exploited from tags and texts, dictionary-encoding structures and delta-encoding geometric coordinate data before the general text compression on back end.We conduct extensive experiments on real GML documents to evaluate the performance of the proposed approach. Results show that our approach outperforms the most popular general text compressor gzip, the acknowledged best XML compressor XMill, and the first and up to now the only GML compressor GPress in compression ratio.
|
[
"Semantic Text Processing",
"Semantic Similarity",
"Information Extraction & Text Mining",
"Text Clustering"
] |
[
72,
53,
3,
29
] |
SCOPUS_ID:77953767491
|
A GOMS model of virtual sociotechnical systems: Using video games to build cognitive models
|
Motivation - The present paper extends the use of GOMS models, described by Kieras (Kieras, 2007) as models of the knowledge necessary for an agent to perform a task, to complex sociotechnical processes involving multiple agents in strategic activities situated in a virtual environment. Research approach - The experiment consists of the SGOMS model, based on task analysis, and a statistical analysis to evaluate the accuracy of the SGOMS model for the description and prediction of the data. Findings/Design - A SGOMS model featuring task interruptions, order violations and planning units representing the decision-making process is a good match with the experimental data. Take away message - The GOMS model can be modified to account for complex sociotechnical interactions within low-fidelity synthetic environments.
|
[
"Visual Data in NLP",
"Cognitive Modeling",
"Linguistics & Cognitive NLP",
"Multimodality"
] |
[
20,
2,
48,
74
] |
SCOPUS_ID:85110867478
|
A GPT-2 language model for biomedical texts in Portuguese
|
Electronic health records (EHRs) contain patient-related information formed by structured and unstructured data, a valuable data source for Natural Language Processing (NLP) in the healthcare domain. The contextual word embeddings and Transformer-based models have proved their potential, reaching state-of-the-art for various NLP tasks. Although the performance for downstream NLP tasks with free-texts written in English has recently improved, less resource is available considering clinical texts and low-resource languages such as Portuguese. Our objective is to develop a Generative Pre-trained Transformer 2 (GPT-2) language model for Portuguese to support clinical and biomedical NLP tasks. We fine-tuned a generic Portuguese GPT-2 model to corpora of biomedical texts written in Portuguese, using transfer learning. We experimented on a public dataset, manually annotated for detecting patient fall, i.e., a classification task. Our in-domain GPT-2 model outperformed the generic Portuguese GPT-2 model by 3.43 in F1-score (weighted). Our preliminary results show that transfer learning with domain literature can benefit Portuguese biomedical NLP tasks, aligned with other languages' results.
|
[
"Language Models",
"Semantic Text Processing"
] |
[
52,
72
] |
SCOPUS_ID:84980411174
|
A GPU-based MapReduce framework for MSR-Bing Image Retrieval Challenge
|
This paper presents a large-scale image retrieval system based on an efficient Graphics Processing Units (GPU)-based MapReduce framework for the MSR-Bing Image Retrieval Challenge. The proposed system is designed for searching images and scoring image-query pairs based on their relevances efficiently and accurately. Unlike the former systems which usually start with text queries to select partial images and then process their visual contents, the proposed system attempts to search similar images directly from the entire dataset through visual content and then compare their text similarities, owing to the powerful computational capabilities of the proposed GPU-based MapReduce framework. It is shown that the proposed system achieves 0.492 in terms of DCG@25 on the final evaluation.
|
[
"Visual Data in NLP",
"Green & Sustainable NLP",
"Responsible & Trustworthy NLP",
"Information Retrieval",
"Multimodality"
] |
[
20,
68,
4,
24,
74
] |
SCOPUS_ID:84859699943
|
A GPU-based accelerator for chinese word segmentation
|
The task of Chinese word segmentation is to split sequence of Chinese characters into tokens so that the Chinese information can be more easily retrieved by web search engine. Due to the dramatic increase in the amount of Chinese literature in recent years, it becomes a big challenge for web search engines to analyze massive Chinese information in time. In this paper, we investigate a new approach to high-performance Chinese information processing. We propose a CPU-GPU collaboration model for Chinese word segmentation. In our novel model, a dictionary-based word segmentation approach is proposed to fit GPU architecture. Three basic word segmentation algorithms are applied to evaluate the performance of this model. In addition, we present several optimization strategies to fully exploit the potential computing power of GPU. Our experimental results show that our model can achieve significant performance speedups up to 3-fold compared with the implementations on CPU. © 2012 Springer-Verlag Berlin Heidelberg.
|
[
"Text Segmentation",
"Syntactic Text Processing"
] |
[
21,
15
] |
SCOPUS_ID:85131255772
|
A GRAPH ATTENTION INTERACTIVE REFINE FRAMEWORK WITH CONTEXTUAL REGULARIZATION FOR JOINTING INTENT DETECTION AND SLOT FILLING
|
Intent detection and slot filling are two important tasks for spoken language understanding. Considering the close relation between them, most existing methods joint them by sharing parameters or establishing explicit connection between them for potentially benefiting each other. However, most of them only consider single directional connection and ignore their cross-impact between them. Moreover, these joint methods treat the predicted labels as the gold labels, which may cause error propagation. In this paper, we propose a two-stage Graph Attention Interactive Refine (GAIR) framework. In stage one, the basic SLU model predicts the coarse intent and slots. In stage two, we select the top-k candidate labels from stage one and construct a graph to make full advantage of intent and slot filling information. By constructing such graph, our framework can establish a bidirectional connection between two tasks and refine the coarse result, which can better take full use of cross-impact between two tasks. Moreover, contextual regularization is introduced for better alleviating error propagation. Experiments on two datasets show that our model achieves the state-of-the-arts performance.
|
[
"Semantic Text Processing",
"Semantic Parsing",
"Structured Data in NLP",
"Intent Recognition",
"Sentiment Analysis",
"Multimodality"
] |
[
72,
40,
50,
79,
78,
74
] |
SCOPUS_ID:84902771083
|
A GROPING VERSUS 'REAL VIOLENCE' IN COLOMBIA: Contrast as a minimisation strategy
|
This article explores discursive contrasts used to minimise a groping in Colombian newspaper forums. Analysis with critical discourse analysis and grounded theory shows that constant talk about 'real' violence in Colombia limits the groping to being seen primarily in contrast with more commonly discussed examples of crime and violence, including the armed conflict, robbery and murder, and sexual abuse. The contrasts, together with other discursive devices, characterise the perpetrator as a normal, hardworking man; suggest that violence was not present in this act; and portray the woman who was groped as different from victims. The discourse in the forum comments places this groping outside of the category of violence rather than considering it as part of a greater picture of gendered violence in Colombia. The contrasts participate in the muting of women's voices about violence against women in Colombia by silencing the victim's perspective and undermining her pursuit of justice. © 2014 Taylor & Francis.
|
[
"Discourse & Pragmatics",
"Semantic Text Processing"
] |
[
71,
72
] |
SCOPUS_ID:85082300845
|
A GRU-Based Neural Machine Translation Followed by Proper Noun Transliteration
|
Neural machine translation has drastically improved the accuracy of machine translation in recent years. The issue of translating out-of-vocabulary proper nouns (OOV-NNP) is still a hindrance to the betterment of machine translation. In this paper, we introduce neural machine translation followed by Proper Noun Transliteration (NMT-NNPT) to address this issue. We explore the idea of transliteration as a post-processing task on the result of neural machine translation using English–Hindi language pair. This approach further improves the translation accuracy and can be used with any language pair.
|
[
"Machine Translation",
"Text Generation",
"Multilinguality"
] |
[
51,
47,
0
] |
http://arxiv.org/abs/1704.08430v2
|
A GRU-Gated Attention Model for Neural Machine Translation
|
Neural machine translation (NMT) heavily relies on an attention network to produce a context vector for each target word prediction. In practice, we find that context vectors for different target words are quite similar to one another and therefore are insufficient in discriminatively predicting target words. The reason for this might be that context vectors produced by the vanilla attention network are just a weighted sum of source representations that are invariant to decoder states. In this paper, we propose a novel GRU-gated attention model (GAtt) for NMT which enhances the degree of discrimination of context vectors by enabling source representations to be sensitive to the partial translation generated by the decoder. GAtt uses a gated recurrent unit (GRU) to combine two types of information: treating a source annotation vector originally produced by the bidirectional encoder as the history state while the corresponding previous decoder state as the input to the GRU. The GRU-combined information forms a new source annotation vector. In this way, we can obtain translation-sensitive source representations which are then feed into the attention network to generate discriminative context vectors. We further propose a variant that regards a source annotation vector as the current input while the previous decoder state as the history. Experiments on NIST Chinese-English translation tasks show that both GAtt-based models achieve significant improvements over the vanilla attentionbased NMT. Further analyses on attention weights and context vectors demonstrate the effectiveness of GAtt in improving the discrimination power of representations and handling the challenging issue of over-translation.
|
[
"Language Models",
"Machine Translation",
"Semantic Text Processing",
"Representation Learning",
"Text Generation",
"Multilinguality"
] |
[
52,
51,
72,
12,
47,
0
] |
SCOPUS_ID:85131357948
|
A GSM Based Assistive Device for Blind, Deaf and Dumb
|
This paper tries to overcome the shortcomings of the recent technology that fails to enhance the communication between physically disabled people by designing an assistive device. This device uses a GSM modem with a SIM card and no smartphones are needed which makes the device affordable. Here, the sender sends the message to the phone number of the disabled person and he/she receives the SMS that gets converted to text and voice message and vibrations to Braille pad using a microcontroller which is easily readable by the disabled person.
|
[
"Speech & Audio in NLP",
"Multimodality"
] |
[
70,
74
] |
SCOPUS_ID:85063269305
|
A Game Theory Approach for Multi-document Summarization
|
In today’s era, information has been growing exponentially on the web, due to which extraction of relevant and concise information has become a challenging task. To overcome the above problem, a fundamental tool known as summarization techniques has been used for understanding and organizing such large datasets. Recently, researchers have been devoting a lot of effort to develop semantics-based models, so as to improve summarization performance. In this paper, a versatile and principled game theory-based multi-document summarization framework integrated with Wikipedia ontology is proposed. The framework exploits the submodularity hidden in underlying ontology and is optimized using the proposed improved algorithm, to enhance the summarization performance. Results of the proposed approach were evaluated with the ROUGE evaluation metric for different multi-document summarization tasks against human-generated summaries and it outperformed DUC, TAC competitors, and other competitive methods.
|
[
"Semantic Text Processing",
"Linguistic Theories",
"Summarization",
"Knowledge Representation",
"Text Generation",
"Linguistics & Cognitive NLP",
"Information Extraction & Text Mining"
] |
[
72,
57,
30,
18,
47,
48,
3
] |
https://aclanthology.org//W15-4720/
|
A Game-Based Setup for Data Collection and Task-Based Evaluation of Uncertain Information Presentation
|
[
"Text Generation"
] |
[
47
] |
|
http://arxiv.org/abs/1606.07711v4
|
A Game-Theoretic Approach to Word Sense Disambiguation
|
This paper presents a new model for word sense disambiguation formulated in terms of evolutionary game theory, where each word to be disambiguated is represented as a node on a graph whose edges represent word relations and senses are represented as classes. The words simultaneously update their class membership preferences according to the senses that neighboring words are likely to choose. We use distributional information to weigh the influence that each word has on the decisions of the others and semantic similarity information to measure the strength of compatibility among the choices. With this information we can formulate the word sense disambiguation problem as a constraint satisfaction problem and solve it using tools derived from game theory, maintaining the textual coherence. The model is based on two ideas: similar words should be assigned to similar classes and the meaning of a word does not depend on all the words in a text but just on some of them. The paper provides an in-depth motivation of the idea of modeling the word sense disambiguation problem in terms of game theory, which is illustrated by an example. The conclusion presents an extensive analysis on the combination of similarity measures to use in the framework and a comparison with state-of-the-art systems. The results show that our model outperforms state-of-the-art algorithms and can be applied to different tasks and in different scenarios.
|
[
"Linguistics & Cognitive NLP",
"Semantic Text Processing",
"Word Sense Disambiguation",
"Linguistic Theories"
] |
[
48,
72,
65,
57
] |
http://arxiv.org/abs/2101.03269v1
|
A Gamification of Japanese Dependency Parsing
|
Gamification approaches have been used as a way for creating language resources for NLP. It is also used for presenting and teaching the algorithms in NLP and linguistic phenomena. This paper argues about a design of gamification for Japanese syntactic dependendency parsing for the latter objective. The user interface design is based on a transition-based shift reduce dependency parsing which needs only two actions of SHIFT (not attach) and REDUCE (attach) in Japanese dependency structure. We assign the two actions for two-way directional control on a gamepad or other devices. We also design the target sentences from psycholinguistics researches.
|
[
"Syntactic Parsing",
"Syntactic Text Processing"
] |
[
28,
15
] |
SCOPUS_ID:85125296273
|
A Gamified Approach to Automatically Detect Biased Wording and Train Critical Reading
|
Biased media has an effect on the public perception of occurring events. By altering word choice, outlets can alter beliefs and views. A gold standard data set is needed to train sufficient classifiers that detect biased wording. This work aims to develop a game that trains players to read news critically while collecting their annotations. The vision is to tackle the complex problem of media bias detection with a very scalable, high quality, and gold standard data set to overcome the drawbacks of current models in the area.
|
[
"Ethical NLP",
"Responsible & Trustworthy NLP"
] |
[
17,
4
] |
SCOPUS_ID:85051071071
|
A Gamified Approach to Naïve Bayes Classification: A Case Study for Newswires and Systematic Medical Reviews
|
Supervised machine learning algorithms require a set of labelled examples to be trained; however, the labelling process is a costly and time consuming task which is carried out by experts of the domain who label the dataset by means of an iterative process to filter out non-relevant objects of the dataset. In this paper, we describe a set of experiments that use gamification techniques to transform this labelling task into an interactive learning process where users can cooperate in order to achieve a common goal. To this end, first we use a geometrical interpretation of Naïve Bayes (NB) classifiers in order to create an intuitive visualization of the current state of the system and let the user change some of the parameters directly as part of a game. We apply this visualization technique to the classification of newswire and we report the results of the experiments conducted with different groups of people: PhD students, Master Degree students and general public. Then, we present a preliminary experiment of query rewriting for systematic reviews in a medical scenario, which makes use of gamification techniques to collect different formulation of the same query. Both the experiments show how the exploitation of gamification approaches help to engage the users in abstract tasks that might be hard to understand and/or boring to perform.
|
[
"Information Retrieval",
"Text Classification",
"Information Extraction & Text Mining"
] |
[
24,
36,
3
] |
http://arxiv.org/abs/2004.11464v1
|
A Gamma-Poisson Mixture Topic Model for Short Text
|
Most topic models are constructed under the assumption that documents follow a multinomial distribution. The Poisson distribution is an alternative distribution to describe the probability of count data. For topic modelling, the Poisson distribution describes the number of occurrences of a word in documents of fixed length. The Poisson distribution has been successfully applied in text classification, but its application to topic modelling is not well documented, specifically in the context of a generative probabilistic model. Furthermore, the few Poisson topic models in literature are admixture models, making the assumption that a document is generated from a mixture of topics. In this study, we focus on short text. Many studies have shown that the simpler assumption of a mixture model fits short text better. With mixture models, as opposed to admixture models, the generative assumption is that a document is generated from a single topic. One topic model, which makes this one-topic-per-document assumption, is the Dirichlet-multinomial mixture model. The main contributions of this work are a new Gamma-Poisson mixture model, as well as a collapsed Gibbs sampler for the model. The benefit of the collapsed Gibbs sampler derivation is that the model is able to automatically select the number of topics contained in the corpus. The results show that the Gamma-Poisson mixture model performs better than the Dirichlet-multinomial mixture model at selecting the number of topics in labelled corpora. Furthermore, the Gamma-Poisson mixture produces better topic coherence scores than the Dirichlet-multinomial mixture model, thus making it a viable option for the challenging task of topic modelling of short text.
|
[
"Topic Modeling",
"Information Extraction & Text Mining"
] |
[
9,
3
] |
http://arxiv.org/abs/1712.09509v1
|
A Gap-Based Framework for Chinese Word Segmentation via Very Deep Convolutional Networks
|
Most previous approaches to Chinese word segmentation can be roughly classified into character-based and word-based methods. The former regards this task as a sequence-labeling problem, while the latter directly segments character sequence into words. However, if we consider segmenting a given sentence, the most intuitive idea is to predict whether to segment for each gap between two consecutive characters, which in comparison makes previous approaches seem too complex. Therefore, in this paper, we propose a gap-based framework to implement this intuitive idea. Moreover, very deep convolutional neural networks, namely, ResNets and DenseNets, are exploited in our experiments. Results show that our approach outperforms the best character-based and word-based methods on 5 benchmarks, without any further post-processing module (e.g. Conditional Random Fields) nor beam search.
|
[
"Text Segmentation",
"Syntactic Text Processing"
] |
[
21,
15
] |
SCOPUS_ID:85137577544
|
A Gaze into the Internal Logic of Graph Neural Networks, with Logic
|
Graph Neural Networks share with Logic Programming several key relational inference mechanisms. The datasets on which they are trained and evaluated can be seen as database facts containing ground terms. This makes possible modeling their inference mechanisms with equivalent logic programs, to better understand not just how they propagate information between the entities involved in the machine learning process but also to infer limits on what can be learned from a given dataset and how well that might generalize to unseen test data. This leads us to the key idea of this paper: modeling with the help of a logic program the information flows involved in learning to infer from the link structure of a graph and the information content of its nodes properties of new nodes, given their known connections to nodes with possibly similar properties. The problem is known as graph node property prediction and our approach will consist in emulating with help of a Prolog program the key information propagation steps of a Graph Neural Network’s training and inference stages. We test our a approach on the ogbn-arxiv node property inference benchmark. To infer class labels for nodes representing papers in a citation network, we distill the dependency trees of the text associated to each node into directed acyclic graphs that we encode as ground Prolog terms. Together with the set of their references to other papers, they become facts in a database on which we reason with help of a Prolog program that mimics the information propagation in graph neural networks predicting node properties. In the process, we invent ground term similarity relations that help infer labels in the test set by propagating node properties from similar nodes in the training set and we evaluate their effectiveness in comparison with that of the graph’s link structure. Finally, we implement explanation generators that unveil performance upper bounds inherent to the dataset. As a practical outcome, we obtain a logic program, that, when seen as machine learning algorithm, performs close to the state of the art on the node property prediction benchmark.
|
[
"Programming Languages in NLP",
"Structured Data in NLP",
"Multimodality"
] |
[
55,
50,
74
] |
SCOPUS_ID:0021386405
|
A General Approach to Inference of Context-Free Programmed Grammars
|
A general approach to the context-free programmed grammars (CFPG) inference is proposed on the basis of inferability analysis. The method is applicable to a sufficiently large class of languages for a string pattern description in syntactic pattern recognition. Especially important is that languages with basic recursive structures whose recursive parameter can be represented by polynomial functions with finite terms are always suitable for the suggested method. Five major problems have been discussed: sample set generalization, string segmentation, CFPG trunk grammar inference, CFPG subgrammar inference, and derivation program unification. © 1984 IEEE
|
[
"Text Error Correction",
"Syntactic Text Processing",
"Programming Languages in NLP",
"Multimodality"
] |
[
26,
15,
55,
74
] |
https://aclanthology.org//2020.webnlg-1.3/
|
A General Benchmarking Framework for Text Generation
|
The RDF-to-text task has recently gained substantial attention due to the continuous growth of RDF knowledge graphs in number and size. Recent studies have focused on systematically comparing RDF-to-text approaches on benchmarking datasets such as WebNLG. Although some evaluation tools have already been proposed for text generation, none of the existing solutions abides by the Findability, Accessibility, Interoperability, and Reusability (FAIR) principles and involves RDF data for the knowledge extraction task. In this paper, we present BENG, a FAIR benchmarking platform for Natural Language Generation (NLG) and Knowledge Extraction systems with focus on RDF data. BENG builds upon the successful benchmarking platform GERBIL, is opensource and is publicly available along with the data it contains.
|
[
"Text Generation",
"Information Extraction & Text Mining"
] |
[
47,
3
] |
http://arxiv.org/abs/2207.05948v1
|
A General Contextualized Rewriting Framework for Text Summarization
|
The rewriting method for text summarization combines extractive and abstractive approaches, improving the conciseness and readability of extractive summaries using an abstractive model. Exiting rewriting systems take each extractive sentence as the only input, which is relatively focused but can lose necessary background knowledge and discourse context. In this paper, we investigate contextualized rewriting, which consumes the entire document and considers the summary context. We formalize contextualized rewriting as a seq2seq with group-tag alignments, introducing group-tag as a solution to model the alignments, identifying extractive sentences through content-based addressing. Results show that our approach significantly outperforms non-contextualized rewriting systems without requiring reinforcement learning, achieving strong improvements on ROUGE scores upon multiple extractors.
|
[
"Summarization",
"Paraphrasing",
"Text Generation",
"Information Extraction & Text Mining"
] |
[
30,
32,
47,
3
] |
SCOPUS_ID:85056622378
|
A General Critical Discourse Analysis Framework for Educational Research
|
Critical discourse analysis (CDA) is a qualitative analytical approach for critically describing, interpreting, and explaining the ways in which discourses construct, maintain, and legitimize social inequalities. CDA rests on the notion that the way we use language is purposeful, regardless of whether discursive choices are conscious or unconscious. CDA takes a number of different approaches and incorporates a variety of methods that depend on research goals and theoretical perspectives. This methodological guide presents a general CDA analytic framework and illustrates the application of that framework to a systematic literature review of CDA studies in education. CDA research studies are no less likely than other forms of scholarly research to reproduce ideological assumptions; qualitative rigor and trustworthiness are discussed.
|
[
"Discourse & Pragmatics",
"Semantic Text Processing"
] |
[
71,
72
] |
http://arxiv.org/abs/1903.12356v1
|
A General FOFE-net Framework for Simple and Effective Question Answering over Knowledge Bases
|
Question answering over knowledge base (KB-QA) has recently become a popular research topic in NLP. One popular way to solve the KB-QA problem is to make use of a pipeline of several NLP modules, including entity discovery and linking (EDL) and relation detection. Recent success on KB-QA task usually involves complex network structures with sophisticated heuristics. Inspired by a previous work that builds a strong KB-QA baseline, we propose a simple but general neural model composed of fixed-size ordinally forgetting encoding (FOFE) and deep neural networks, called FOFE-net to solve KB-QA problem at different stages. For evaluation, we use two popular KB-QA datasets, SimpleQuestions and WebQSP, and a newly created dataset, FreebaseQA. The experimental results show that FOFE-net performs well on KB-QA subtasks, entity discovery and linking (EDL) and relation detection, and in turn pushing overall KB-QA system to achieve strong results on all datasets.
|
[
"Semantic Text Processing",
"Question Answering",
"Knowledge Representation",
"Named Entity Recognition",
"Natural Language Interfaces",
"Information Extraction & Text Mining"
] |
[
72,
27,
18,
34,
11,
3
] |
http://arxiv.org/abs/1911.03154v2
|
A General Framework for Adaptation of Neural Machine Translation to Simultaneous Translation
|
Despite the success of neural machine translation (NMT), simultaneous neural machine translation (SNMT), the task of translating in real time before a full sentence has been observed, remains challenging due to the syntactic structure difference and simultaneity requirements. In this paper, we propose a general framework for adapting neural machine translation to translate simultaneously. Our framework contains two parts: prefix translation that utilizes a consecutive NMT model to translate source prefixes and a stopping criterion that determines when to stop the prefix translation. Experiments on three translation corpora and two language pairs show the efficacy of the proposed framework on balancing the quality and latency in adapting NMT to perform simultaneous translation.
|
[
"Machine Translation",
"Text Generation",
"Multilinguality"
] |
[
51,
47,
0
] |
http://arxiv.org/abs/1610.02906v3
|
A General Framework for Content-enhanced Network Representation Learning
|
This paper investigates the problem of network embedding, which aims at learning low-dimensional vector representation of nodes in networks. Most existing network embedding methods rely solely on the network structure, i.e., the linkage relationships between nodes, but ignore the rich content information associated with it, which is common in real world networks and beneficial to describing the characteristics of a node. In this paper, we propose content-enhanced network embedding (CENE), which is capable of jointly leveraging the network structure and the content information. Our approach integrates text modeling and structure modeling in a general framework by treating the content information as a special kind of node. Experiments on several real world net- works with application to node classification show that our models outperform all existing network embedding methods, demonstrating the merits of content information and joint learning.
|
[
"Semantic Text Processing",
"Representation Learning"
] |
[
72,
12
] |
http://arxiv.org/abs/2111.14309v1
|
A General Framework for Defending Against Backdoor Attacks via Influence Graph
|
In this work, we propose a new and general framework to defend against backdoor attacks, inspired by the fact that attack triggers usually follow a \textsc{specific} type of attacking pattern, and therefore, poisoned training examples have greater impacts on each other during training. We introduce the notion of the {\it influence graph}, which consists of nodes and edges respectively representative of individual training points and associated pair-wise influences. The influence between a pair of training points represents the impact of removing one training point on the prediction of another, approximated by the influence function \citep{koh2017understanding}. Malicious training points are extracted by finding the maximum average sub-graph subject to a particular size. Extensive experiments on computer vision and natural language processing tasks demonstrate the effectiveness and generality of the proposed framework.
|
[
"Responsible & Trustworthy NLP",
"Structured Data in NLP",
"Robustness in NLP",
"Multimodality"
] |
[
4,
50,
58,
74
] |
SCOPUS_ID:85116857883
|
A General Framework for First Story Detection Utilizing Entities and Their Relations
|
News portals, such as Yahoo News or Google News, collect large amounts of news articles from a variety of sources on a daily basis. Only a small portion of these documents can be selected and displayed on the homepage. Thus, there is a strong preference for major, recent events. In this work, we propose a scalable First Story Detection (FSD) pipeline that identifies fresh news. This pipeline is used in order to instantiate a variety of FSD approaches. In addition we suggest a novel FSD technique that in comparison to existing systems, relies on relation extraction algorithms and exploits the named entities and their relations in order to decide about the freshness of an article. We evaluate our technique by instantiating existing state of art FSD techniques within our generic pipeline. As ground truth we use multiple datasets that cover different categories. Experimental results demonstrate that our FSD method in many cases provides an improvement over state-of-the-art techniques. In addition, we show using a large synthetic dataset that our general FSD pipeline has constant space and time requirements and is suitable for very high volume streams.
|
[
"Event Extraction",
"Information Extraction & Text Mining"
] |
[
31,
3
] |
http://arxiv.org/abs/1909.06092v2
|
A General Framework for Implicit and Explicit Debiasing of Distributional Word Vector Spaces
|
Distributional word vectors have recently been shown to encode many of the human biases, most notably gender and racial biases, and models for attenuating such biases have consequently been proposed. However, existing models and studies (1) operate on under-specified and mutually differing bias definitions, (2) are tailored for a particular bias (e.g., gender bias) and (3) have been evaluated inconsistently and non-rigorously. In this work, we introduce a general framework for debiasing word embeddings. We operationalize the definition of a bias by discerning two types of bias specification: explicit and implicit. We then propose three debiasing models that operate on explicit or implicit bias specifications and that can be composed towards more robust debiasing. Finally, we devise a full-fledged evaluation framework in which we couple existing bias metrics with newly proposed ones. Experimental findings across three embedding methods suggest that the proposed debiasing models are robust and widely applicable: they often completely remove the bias both implicitly and explicitly without degradation of semantic information encoded in any of the input distributional spaces. Moreover, we successfully transfer debiasing models, by means of cross-lingual embedding spaces, and remove or attenuate biases in distributional word vector spaces of languages that lack readily available bias specifications.
|
[
"Responsible & Trustworthy NLP",
"Semantic Text Processing",
"Robustness in NLP",
"Representation Learning"
] |
[
4,
72,
58,
12
] |
http://arxiv.org/abs/1904.03296v1
|
A General Framework for Information Extraction using Dynamic Span Graphs
|
We introduce a general framework for several information extraction tasks that share span representations using dynamically constructed span graphs. The graphs are constructed by selecting the most confident entity spans and linking these nodes with confidence-weighted relation types and coreferences. The dynamic span graph allows coreference and relation type confidences to propagate through the graph to iteratively refine the span representations. This is unlike previous multi-task frameworks for information extraction in which the only interaction between tasks is in the shared first-layer LSTM. Our framework significantly outperforms the state-of-the-art on multiple information extraction tasks across multiple datasets reflecting different domains. We further observe that the span enumeration approach is good at detecting nested span entities, with significant F1 score improvement on the ACE dataset.
|
[
"Multimodality",
"Structured Data in NLP",
"Coreference Resolution",
"Information Extraction & Text Mining"
] |
[
74,
50,
13,
3
] |
http://arxiv.org/abs/2103.12615v1
|
A General Framework for Learning Prosodic-Enhanced Representation of Rap Lyrics
|
Learning and analyzing rap lyrics is a significant basis for many web applications, such as music recommendation, automatic music categorization, and music information retrieval, due to the abundant source of digital music in the World Wide Web. Although numerous studies have explored the topic, knowledge in this field is far from satisfactory, because critical issues, such as prosodic information and its effective representation, as well as appropriate integration of various features, are usually ignored. In this paper, we propose a hierarchical attention variational autoencoder framework (HAVAE), which simultaneously consider semantic and prosodic features for rap lyrics representation learning. Specifically, the representation of the prosodic features is encoded by phonetic transcriptions with a novel and effective strategy~(i.e., rhyme2vec). Moreover, a feature aggregation strategy is proposed to appropriately integrate various features and generate prosodic-enhanced representation. A comprehensive empirical evaluation demonstrates that the proposed framework outperforms the state-of-the-art approaches under various metrics in different rap lyrics learning tasks.
|
[
"Representation Learning",
"Semantic Text Processing",
"Speech & Audio in NLP",
"Multimodality"
] |
[
12,
72,
70,
74
] |
SCOPUS_ID:85074604960
|
A General Framework for Multiple Choice Question Answering Based on Mutual Information and Reinforced Co-occurrence
|
As a result of the continuously growing volume of information available, browsing and querying of textual information in search of specific facts is currently a tedious task exacerbated by a reality where data presentation very often does not meet the needs of users. To satisfy these ever-increasing needs, we have designed an solution to provide an adaptive and intelligent solution for the automatic answer of multiple-choice questions based on the concept of mutual information. An empirical evaluation over a number of general-purpose benchmark datasets seems to indicate that this solution is promising.
|
[
"Natural Language Interfaces",
"Question Answering"
] |
[
11,
27
] |
SCOPUS_ID:84936758447
|
A General Hospital and its conceptions of madness
|
The aim of this article was to survey the conceptions of madness produced by professionals working in a general hospital. Procedures: conversation groups were conducted and the results were analyzed based on Discourse Analysis and Michel Foucault’s History of Madness. Conclusions: the structures of the asylums remain installed in their concreteness and in the mentality of people, who continue to believe that people suffering from madness must be controlled to meet social demands.
|
[
"Discourse & Pragmatics",
"Semantic Text Processing"
] |
[
71,
72
] |
SCOPUS_ID:85148233661
|
A General Linguistic Steganalysis Framework Using Multi-Task Learning
|
Prevailing linguistic steganalysis approaches focus on learning sensitive features to distinguish a particular category of steganographic texts from non-steganographic texts, by performing binary classification. While it remains an unsolved problem and poses a significant threat to the security of cyberspace when various categories of non-steganographic or steganographic texts coexist. In this paper, we propose a general linguistic steganalysis framework named LS-MTL, which introduces the idea of multi-task learning to deal with the classification of various categories of steganographic and non-steganographic texts. LS-MTL captures sensitive linguistic features from multiple related linguistic steganalysis tasks and can concurrently handle diverse tasks with a constructed model. In the proposed framework, convolutional neural networks (CNNs) are utilized as private base models to extract sensitive features for each steganalysis task. Besides, a shared CNN is built to capture potential interaction information and share linguistic features among all tasks. Finally, LS-MTL incorporates the private and shared sensitive features to identify the detected text as steganographic or non-steganographic. Experimental results demonstrate that the proposed framework LS-MTL outperforms the baseline in the multi-category linguistic steganalysis task, while average Acc, Pre, and Rec are increased by 0.5%, 1.4%, and 0.4%, respectively. More ablation experimental results show that LS-MTL with the shared module has robust generalization capability and achieves good detection performance even in the case of spare data.
|
[
"Low-Resource NLP",
"Language Models",
"Semantic Text Processing",
"Information Retrieval",
"Ethical NLP",
"Responsible & Trustworthy NLP",
"Text Classification",
"Information Extraction & Text Mining"
] |
[
80,
52,
72,
24,
17,
4,
36,
3
] |
SCOPUS_ID:85117367775
|
A General Method for Transferring Explicit Knowledge into Language Model Pretraining
|
Recently, pretrained language models, such as Bert and XLNet, have rapidly advanced the state of the art on many NLP tasks. They can model implicit semantic information between words in the text. However, it is solely at the token level without considering the background knowledge. Intuitively, background knowledge influences the efficacy of text understanding. Inspired by this, we focus on improving model pretraining by leveraging external knowledge. Different from recent research that optimizes pretraining models by knowledge masking strategies, we propose a simple but general method to transfer explicit knowledge with pretraining. To be specific, we first match knowledge facts from a knowledge base (KB) and then add a knowledge injunction layer to a transformer directly without changing its architecture. This study seeks to find the direct impact of explicit knowledge on model pretraining. We conduct experiments on 7 datasets using 5 knowledge bases in different downstream tasks. Our investigation reveals promising results in all the tasks. The experiment also verifies that domain-specific knowledge is superior to open-domain knowledge in domain-specific task, and different knowledge bases have different performances in different tasks.
|
[
"Language Models",
"Knowledge Representation",
"Semantic Text Processing"
] |
[
52,
18,
72
] |
http://arxiv.org/abs/2010.11338v2
|
A General Multi-Task Learning Framework to Leverage Text Data for Speech to Text Tasks
|
Attention-based sequence-to-sequence modeling provides a powerful and elegant solution for applications that need to map one sequence to a different sequence. Its success heavily relies on the availability of large amounts of training data. This presents a challenge for speech applications where labelled speech data is very expensive to obtain, such as automatic speech recognition (ASR) and speech translation (ST). In this study, we propose a general multi-task learning framework to leverage text data for ASR and ST tasks. Two auxiliary tasks, a denoising autoencoder task and machine translation task, are proposed to be co-trained with ASR and ST tasks respectively. We demonstrate that representing text input as phoneme sequences can reduce the difference between speech and text inputs, and enhance the knowledge transfer from text corpora to the speech to text tasks. Our experiments show that the proposed method achieves a relative 10~15% word error rate reduction on the English Librispeech task compared with our baseline, and improves the speech translation quality on the MuST-C tasks by 3.6~9.2 BLEU.
|
[
"Multilinguality",
"Language Models",
"Low-Resource NLP",
"Machine Translation",
"Semantic Text Processing",
"Speech & Audio in NLP",
"Multimodality",
"Text Generation",
"Speech Recognition",
"Responsible & Trustworthy NLP"
] |
[
0,
52,
80,
51,
72,
70,
74,
47,
10,
4
] |
SCOPUS_ID:85026675326
|
A General Multimedia Representation Space Model toward Event-Based Collective Knowledge Management
|
Emergent technologies such as smart phones and wireless Internet have transformed the Web from a static data publishing platform into a collaborative information sharing environment. Yet, attaining the next stage in Web engineering, i.e., the so-called Intelligent Web: allowing meaningful human-machine and machine-machine collaboration, requires another breakthrough: allowing the sharing and organization of collective knowledge (CK), where CK underlines the combination of all known data, information, and meta-data concerning a given concept or event. In this context, various methods have been put forward to perform automatic event extraction and description. Yet, most of them do not capture the semantic meaning embedded in Web-based multimedia data, which are usually highly heterogeneous and unstructured. To address this problem, we introduce in this study a generic Multimedia Representation Space Model called MRSM, designed for multimedia data and multimedia-based event representation, in order to allow event detection and identification based on multimedia CK. We formally define MRSM, its dimensions, their coordinates, and the associated distance (similarity) metrics and properties. We then provide the building blocks for an Eventbased Collective Knowledge (CK) Management Framework, built upon MRSM, and geared toward effective CK management. The proposed approach provides a means of extracting, representing, and linking events from heterogeneous multimedia data without any prior knowledge about event-related clues. Preliminary tests confirm the quality and potential of our approach.
|
[
"Event Extraction",
"Information Extraction & Text Mining",
"Semantic Text Processing",
"Representation Learning"
] |
[
31,
3,
72,
12
] |
SCOPUS_ID:85140765034
|
A General Multiple Data Augmentation Based Framework for Training Deep Neural Networks
|
Deep neural networks (DNNs) often rely on massive labelled data for training, which is inaccessible in many applications. Data augmentation (DA) tackles data scarcity by creating new labelled data from available ones. Different DA methods have different mechanisms and therefore using their generated labelled data for DNN training may help improving DNN's generalisation to different degrees. Combining multiple DA methods, namely multi-DA, for DNN training, provides a way to further boost generalisation. Among existing multi-DA based DNN training methods, those relying on knowledge distillation (KD) have received great attention. They leverage knowledge transfer to utilise the labelled data sets created by multiple DA methods instead of directly combining them for training DNNs. However, existing KD-based methods can only utilise certain types of DA methods, incapable of making full use of the advantages of arbitrary DA methods. In this work, we propose a general multi-DA based DNN training framework capable to use arbitrary DA methods. To train a DNN, our framework replicates a certain portion in the latter part of the DNN into multiple copies, leading to multiple DNNs with shared blocks in their former parts and independent blocks in their latter parts. Each of these DNNs is associated with a unique DA and a newly devised loss that allows comprehensively learning from the data generated by all DA methods and the outputs from all DNNs in an online and adaptive way. The overall loss, i.e., the sum of each DNN's loss, is used for training the DNN. Eventually, one of the DNNs with the best validation performance is chosen for inference. We implement the proposed framework by using three distinct DA methods and apply it for training representative DNNs. Experimental results on the popular benchmarks of image classification demonstrate the superiority of our method to several existing single-DA and multi-DA based training methods.
|
[
"Low-Resource NLP",
"Responsible & Trustworthy NLP",
"Green & Sustainable NLP"
] |
[
80,
4,
68
] |
SCOPUS_ID:85083178538
|
A General Procedure for Improving Language Models in Low-Resource Speech Recognition
|
It is difficult for a language model (LM) to perform well with limited in-domain transcripts in low-resource speech recognition. In this paper, we mainly summarize and extend some effective methods to make the most of the out-of-domain data to improve LMs. These methods include data selection, vocabulary expansion, lexicon augmentation, multi-model fusion and so on. The methods are integrated into a systematic procedure, which proves to be effective for improving both n-gram and neural network LMs. Additionally, pre-Trained word vectors using out-of-domain data are utilized to improve the performance of RNN/LSTM LMs for rescoring first-pass decoding results. Experiments on five Asian languages from Babel Build Packs show that, after improving LMs, 5.4-7.6% relative reduction of word error rate (WER) is generally achieved compared to the baseline ASR systems. For some languages, we achieve lower WER than newly published results on the same data sets.
|
[
"Language Models",
"Low-Resource NLP",
"Semantic Text Processing",
"Speech & Audio in NLP",
"Multimodality",
"Text Generation",
"Speech Recognition",
"Responsible & Trustworthy NLP"
] |
[
52,
80,
72,
70,
74,
47,
10,
4
] |
SCOPUS_ID:85066117874
|
A General Process for the Semantic Annotation and Enrichment of Electronic Program Guides
|
Electronic Program Guides (EPGs) are usual resources aimed to inform the audience about the programming being transmitted by TV stations and cable/satellite TV providers. However, they only provide basic metadata about the TV programs, while users may want to obtain additional information related to the content they are currently watching. This paper proposes a general process for the semantic annotation and subsequent enrichment of EPGs using external knowledge bases and natural language processing techniques with the aim to tackle the lack of immediate availability of related information about TV programs. Additionally, we define an evaluation approach based on a distributed representation of words that can enable TV content providers to verify the effectiveness of the system and perform an automatic execution of the enrichment process. We test our proposal using a real-world dataset and demonstrate its effectiveness by using different knowledge bases, word representation models and similarity measures. Results showed that DBpedia and Google Knowledge Graph knowledge bases return the most relevant content during the enrichment process, while word2vec and fasttext models with Words Mover’s Distance as similarity function can be combined to validate the effectiveness of the retrieval task.
|
[
"Programming Languages in NLP",
"Semantic Text Processing",
"Representation Learning",
"Knowledge Representation",
"Multimodality"
] |
[
55,
72,
12,
18,
74
] |
SCOPUS_ID:85139592245
|
A General Purpose Turkish CLIP Model (TrCLIP) for Image&Text Retrieval and its Application to E-Commerce
|
In this paper, we introduce a Turkish adaption of CLIP (Contrastive Language-Image Pre-Training). Our approach is to train a model with the same output space as the Text encoder of the CLIP model while processing Turkish input. For this, we collected 2.5M unique English-Turkish data. The model we named TrCLIP performed 71% in CIFAR100, 86% in VOC2007, and 47% in FER2013 as zero-shot accuracy. We have examined its performance on e-commerce data and a vast domain-independent dataset in image and text retrieval tasks. The model can work in Turkish without any extra fine-tuning. Models and dataset can be reachable from https://github.com/yusufani/TrCLIP.
|
[
"Visual Data in NLP",
"Low-Resource NLP",
"Multimodality",
"Information Retrieval",
"Responsible & Trustworthy NLP"
] |
[
20,
80,
74,
24,
4
] |
http://arxiv.org/abs/cmp-lg/9801005v1
|
A General, Sound and Efficient Natural Language Parsing Algorithm based on Syntactic Constraints Propagation
|
This paper presents a new context-free parsing algorithm based on a bidirectional strictly horizontal strategy which incorporates strong top-down predictions (derivations and adjacencies). From a functional point of view, the parser is able to propagate syntactic constraints reducing parsing ambiguity. From a computational perspective, the algorithm includes different techniques aimed at the improvement of the manipulation and representation of the structures used.
|
[
"Responsible & Trustworthy NLP",
"Syntactic Text Processing",
"Green & Sustainable NLP"
] |
[
4,
15,
68
] |
https://aclanthology.org//W11-1015/
|
A General-Purpose Rule Extractor for SCFG-Based Machine Translation
|
[
"Machine Translation",
"Text Generation",
"Multilinguality"
] |
[
51,
47,
0
] |
|
http://arxiv.org/abs/1602.01635v2
|
A Generalised Quantifier Theory of Natural Language in Categorical Compositional Distributional Semantics with Bialgebras
|
Categorical compositional distributional semantics is a model of natural language; it combines the statistical vector space models of words with the compositional models of grammar. We formalise in this model the generalised quantifier theory of natural language, due to Barwise and Cooper. The underlying setting is a compact closed category with bialgebras. We start from a generative grammar formalisation and develop an abstract categorical compositional semantics for it, then instantiate the abstract setting to sets and relations and to finite dimensional vector spaces and linear maps. We prove the equivalence of the relational instantiation to the truth theoretic semantics of generalised quantifiers. The vector space instantiation formalises the statistical usages of words and enables us to, for the first time, reason about quantified phrases and sentences compositionally in distributional semantics.
|
[
"Semantic Text Processing",
"Linguistic Theories",
"Text Classification",
"Representation Learning",
"Linguistics & Cognitive NLP",
"Information Retrieval",
"Information Extraction & Text Mining"
] |
[
72,
57,
36,
12,
48,
24,
3
] |
SCOPUS_ID:85140793739
|
A Generalized Approach to Protest Event Detection in German Local News
|
Protest events provide information about social and political conflicts, the state of social cohesion and democratic conflict management, as well as the state of civil society in general. Social scientists are therefore interested in the systematic observation of protest events. With this paper, we release the first German language resource of protest event related article excerpts published in local news outlets. We use this dataset to train and evaluate transformer-based text classifiers to automatically detect relevant newspaper articles. Our best approach reaches a binary F1-score of 93.3 %, which is a promising result for our goal to support political science research. However, in a second experiment, we show that our model does not generalize equally well when applied to data from time periods and localities other than our training sample. To make protest event detection more robust, we test two ways of alternative article preprocessing. First, we find that letting the classifier concentrate on sentences around protest keywords only slightly improves the performance for in-sample data. For out-of-sample data, in contrast, binary F1-scores improve up to +4 percentage points (pp). Second, against our initial intuition, masking of named entities during preprocessing does not improve the generalization of protest event detection models in terms of F1-scores. However, it leads to a significantly improved recall of the models.
|
[
"Event Extraction",
"Information Retrieval",
"Text Classification",
"Information Extraction & Text Mining"
] |
[
31,
24,
36,
3
] |
SCOPUS_ID:85034666604
|
A Generalized Constraint Approach to Bilingual Dictionary Induction for Low-Resource Language Families
|
The lack or absence of parallel and comparable corpora makes bilingual lexicon extraction a difficult task for low-resource languages. The pivot language and cognate recognition approaches have been proven useful for inducing bilingual lexicons for such languages. We propose constraint-based bilingual lexicon induction for closely related languages by extending constraints from the recent pivot-based induction technique and further enabling multiple symmetry assumption cycle to reach many more cognates in the transgraph. We further identify cognate synonyms to obtain many-to-many translation pairs. This article utilizes four datasets: one Austronesian low-resource language and three Indo-European high-resource languages. We use three constraint-based methods from our previous work, the Inverse Consultation method and translation pairs generated from Cartesian product of input dictionaries as baselines. We evaluate our result using the metrics of precision, recall, and F-score. Our customizable approach allows the user to conduct cross validation to predict the optimal hyperparameters (cognate threshold and cognate synonym threshold) with various combination of heuristics and number of symmetry assumption cycles to gain the highest F-score. Our proposed methods have statistically significant improvement of precision and F-score compared to our previous constraint-based methods. The results show that our method demonstrates the potential to complement other bilingual dictionary creation methods like word alignment models using parallel corpora for high-resource languages while well handling low-resource languages.
|
[
"Multilinguality",
"Low-Resource NLP",
"Machine Translation",
"Text Generation",
"Responsible & Trustworthy NLP"
] |
[
0,
80,
51,
47,
4
] |
http://arxiv.org/abs/1905.12790v2
|
A Generalized Framework of Sequence Generation with Application to Undirected Sequence Models
|
Undirected neural sequence models such as BERT (Devlin et al., 2019) have received renewed interest due to their success on discriminative natural language understanding tasks such as question-answering and natural language inference. The problem of generating sequences directly from these models has received relatively little attention, in part because generating from undirected models departs significantly from conventional monotonic generation in directed sequence models. We investigate this problem by proposing a generalized model of sequence generation that unifies decoding in directed and undirected models. The proposed framework models the process of generation rather than the resulting sequence, and under this framework, we derive various neural sequence models as special cases, such as autoregressive, semi-autoregressive, and refinement-based non-autoregressive models. This unification enables us to adapt decoding algorithms originally developed for directed sequence models to undirected sequence models. We demonstrate this by evaluating various handcrafted and learned decoding strategies on a BERT-like machine translation model (Lample & Conneau, 2019). The proposed approach achieves constant-time translation results on par with linear-time translation results from the same undirected sequence model, while both are competitive with the state-of-the-art on WMT'14 English-German translation.
|
[
"Language Models",
"Machine Translation",
"Semantic Text Processing",
"Text Generation",
"Multilinguality"
] |
[
52,
51,
72,
47,
0
] |
http://arxiv.org/abs/1404.3377v1
|
A Generalized Language Model as the Combination of Skipped n-grams and Modified Kneser-Ney Smoothing
|
We introduce a novel approach for building language models based on a systematic, recursive exploration of skip n-gram models which are interpolated using modified Kneser-Ney smoothing. Our approach generalizes language models as it contains the classical interpolation with lower order models as a special case. In this paper we motivate, formalize and present our approach. In an extensive empirical experiment over English text corpora we demonstrate that our generalized language models lead to a substantial reduction of perplexity between 3.1% and 12.7% in comparison to traditional language models using modified Kneser-Ney smoothing. Furthermore, we investigate the behaviour over three other languages and a domain specific corpus where we observed consistent improvements. Finally, we also show that the strength of our approach lies in its ability to cope in particular with sparse training data. Using a very small training data set of only 736 KB text we yield improvements of even 25.7% reduction of perplexity.
|
[
"Language Models",
"Semantic Text Processing"
] |
[
52,
72
] |
http://arxiv.org/abs/1901.11167v1
|
A Generalized Language Model in Tensor Space
|
In the literature, tensors have been effectively used for capturing the context information in language models. However, the existing methods usually adopt relatively-low order tensors, which have limited expressive power in modeling language. Developing a higher-order tensor representation is challenging, in terms of deriving an effective solution and showing its generality. In this paper, we propose a language model named Tensor Space Language Model (TSLM), by utilizing tensor networks and tensor decomposition. In TSLM, we build a high-dimensional semantic space constructed by the tensor product of word vectors. Theoretically, we prove that such tensor representation is a generalization of the n-gram language model. We further show that this high-order tensor representation can be decomposed to a recursive calculation of conditional probability for language modeling. The experimental results on Penn Tree Bank (PTB) dataset and WikiText benchmark demonstrate the effectiveness of TSLM.
|
[
"Language Models",
"Semantic Text Processing"
] |
[
52,
72
] |
http://arxiv.org/abs/1707.02892v1
|
A Generalized Recurrent Neural Architecture for Text Classification with Multi-Task Learning
|
Multi-task learning leverages potential correlations among related tasks to extract common features and yield performance gains. However, most previous works only consider simple or weak interactions, thereby failing to model complex correlations among three or more tasks. In this paper, we propose a multi-task learning architecture with four types of recurrent neural layers to fuse information across multiple related tasks. The architecture is structurally flexible and considers various interactions among tasks, which can be regarded as a generalized case of many previous works. Extensive experiments on five benchmark datasets for text classification show that our model can significantly improve performances of related tasks with additional information from others.
|
[
"Language Models",
"Low-Resource NLP",
"Semantic Text Processing",
"Information Retrieval",
"Information Extraction & Text Mining",
"Text Classification",
"Responsible & Trustworthy NLP"
] |
[
52,
80,
72,
24,
3,
36,
4
] |
http://arxiv.org/abs/1807.07779v1
|
A Generalized Vector Space Model for Ontology-Based Information Retrieval
|
Named entities (NE) are objects that are referred to by names such as people, organizations and locations. Named entities and keywords are important to the meaning of a document. We propose a generalized vector space model that combines named entities and keywords. In the model, we take into account different ontological features of named entities, namely, aliases, classes and identifiers. Moreover, we use entity classes to represent the latent information of interrogative words in Wh-queries, which are ignored in traditional keyword-based searching. We have implemented and tested the proposed model on a TREC dataset, as presented and discussed in the paper.
|
[
"Knowledge Representation",
"Semantic Text Processing",
"Information Retrieval",
"Representation Learning"
] |
[
18,
72,
24,
12
] |
https://aclanthology.org//W11-2902/
|
A Generalized View on Parsing and Translation
|
[
"Machine Translation",
"Syntactic Text Processing",
"Syntactic Parsing",
"Text Generation",
"Multilinguality"
] |
[
51,
15,
28,
47,
0
] |
|
https://aclanthology.org//W06-1402/
|
A Generation-Oriented Workbench for Performance Grammar: Capturing Linear Order Variability in German and Dutch
|
[
"Text Generation"
] |
[
47
] |
|
SCOPUS_ID:85138034032
|
A Generative Adversarial Constraint Encoder-Decoder Model for the Text Summarization
|
As a new method of training generative models, Generative Adversarial Net(GAN) has problems when it is applied to the summary generator to generate discrete tokens. This paper considers introducing GAN into the Encoder stage and proposes a new framework EDA(Encoder-Decoder with Adversarial training) for text summarization by combining adversarial training into the traditional Encoder-Decoder architecture. We construct a new training loop and an objective function in this framework to optimize the encoder and decoder. EDA uses GAN to learn the alignment between the encoder representation and the target summary representation to improve the encoding quality and generate more accurate summary by using semantic distance and fine-tuning with decoded summary. It simultaneously alleviates problems of discrete data processing and conditional generation. In particularly, we introduce keyword information of the original document into attention mechanism so that more informative representation and summary can be generated. The results on LCSTS dataset show that our method improves the performance and robustness of Encoder-Decoder model applied in text summarization.
|
[
"Language Models",
"Semantic Text Processing",
"Robustness in NLP",
"Summarization",
"Text Generation",
"Responsible & Trustworthy NLP",
"Information Extraction & Text Mining"
] |
[
52,
72,
58,
30,
47,
4,
3
] |
SCOPUS_ID:85081615010
|
A Generative Adversarial Network Based Ensemble Technique for Automatic Evaluation of Machine Synthesized Speech
|
In this paper, we propose a method to automatically compute a speech evaluation metric, Virtual Mean Opinion Score (vMOS) for the speech generated by Text-to-Speech (TTS) models to analyse its human-ness. In contrast to the currently used manual speech evaluation techniques, the proposed method uses an end-to-end neural network to calculate vMOS which is qualitatively similar to manually obtained Mean Opinion Score (MOS). The Generative Adversarial Network (GAN) and a binary classifier have been trained on real natural speech with known MOS. Further, the vMOS has been calculated by averaging the scores obtained by the two networks. In this work, the input to GAN’s discriminator is conditioned with the speech generated by off-the-shelf TTS models so as to get closer to the natural speech. It has been shown that the proposed model can be trained with a minimum amount of data as its objective is to generate only the evaluation score and not speech. The proposed method has been tested to evaluate the speech synthesized by state-of-the-art TTS models and it has reported the vMOS of 0.6675, 0.4945 and 0.4890 for Wavenet2, Tacotron and Deepvoice3 respectively while the vMOS for natural speech is 0.6682 on a scale from 0 to 1. These vMOS scores correspond to and are qualitatively explained by their manually calculated MOS scores.
|
[
"Responsible & Trustworthy NLP",
"Speech & Audio in NLP",
"Robustness in NLP",
"Multimodality"
] |
[
4,
70,
58,
74
] |
http://arxiv.org/abs/2204.05674v1
|
A Generative Approach for Financial Causality Extraction
|
Causality represents the foremost relation between events in financial documents such as financial news articles, financial reports. Each financial causality contains a cause span and an effect span. Previous works proposed sequence labeling approaches to solve this task. But sequence labeling models find it difficult to extract multiple causalities and overlapping causalities from the text segments. In this paper, we explore a generative approach for causality extraction using the encoder-decoder framework and pointer networks. We use a causality dataset from the financial domain, \textit{FinCausal}, for our experiments and our proposed framework achieves very competitive performance on this dataset.
|
[
"Information Extraction & Text Mining"
] |
[
3
] |
http://arxiv.org/abs/2108.14006v1
|
A Generative Approach for Mitigating Structural Biases in Natural Language Inference
|
Many natural language inference (NLI) datasets contain biases that allow models to perform well by only using a biased subset of the input, without considering the remainder features. For instance, models are able to make a classification decision by only using the hypothesis, without learning the true relationship between it and the premise. These structural biases lead discriminative models to learn unintended superficial features and to generalize poorly out of the training distribution. In this work, we reformulate the NLI task as a generative task, where a model is conditioned on the biased subset of the input and the label and generates the remaining subset of the input. We show that by imposing a uniform prior, we obtain a provably unbiased model. Through synthetic experiments, we find that this approach is highly robust to large amounts of bias. We then demonstrate empirically on two types of natural bias that this approach leads to fully unbiased models in practice. However, we find that generative models are difficult to train and they generally perform worse than discriminative baselines. We highlight the difficulty of the generative modeling task in the context of NLI as a cause for this worse performance. Finally, by fine-tuning the generative model with a discriminative objective, we reduce the performance gap between the generative model and the discriminative baseline, while allowing for a small amount of bias.
|
[
"Reasoning",
"Textual Inference"
] |
[
8,
22
] |
http://arxiv.org/abs/1711.06238v2
|
A Generative Approach to Question Answering
|
Question Answering has come a long way from answer sentence selection, relational QA to reading and comprehension. We shift our attention to generative question answering (gQA) by which we facilitate machine to read passages and answer questions by learning to generate the answers. We frame the problem as a generative task where the encoder being a network that models the relationship between question and passage and encoding them to a vector thus facilitating the decoder to directly form an abstraction of the answer. Not being able to retain facts and making repetitions are common mistakes that affect the overall legibility of answers. To counter these issues, we employ copying mechanism and maintenance of coverage vector in our model respectively. Our results on MS-MARCO demonstrate it's superiority over baselines and we also show qualitative examples where we improved in terms of correctness and readability
|
[
"Natural Language Interfaces",
"Question Answering"
] |
[
11,
27
] |
https://aclanthology.org//2020.ngt-1.9/
|
A Generative Approach to Titling and Clustering Wikipedia Sections
|
We evaluate the performance of transformer encoders with various decoders for information organization through a new task: generation of section headings for Wikipedia articles. Our analysis shows that decoders containing attention mechanisms over the encoder output achieve high-scoring results by generating extractive text. In contrast, a decoder without attention better facilitates semantic encoding and can be used to generate section embeddings. We additionally introduce a new loss function, which further encourages the decoder to generate high-quality embeddings.
|
[
"Language Models",
"Semantic Text Processing",
"Representation Learning",
"Text Generation",
"Text Clustering",
"Information Extraction & Text Mining"
] |
[
52,
72,
12,
47,
29,
3
] |
SCOPUS_ID:85128251238
|
A Generative Approach to the Instructed Second Language Acquisition of Spanish se
|
This article focuses on the role of crosslinguistic patterns with verbs in the mapping of noun phrases/semantic roles to positions in morphosyntax, with a particular focus on second language (L2) development of Spanish se. The data set derives from high school learners of Spanish in the United States under broadly deductive and inductive learning treatments leading to explicit awareness. Using linear mixed effects modeling (LME) and binomial logistic regression, an analysis of high school learners from three schools (total n = 138) showed that learners based their acceptability judgments of aurally presented sentences and written production on verb classes proposed in formal linguistic theory. However, effects of the instructional intervention were limited to production data. No advantage for either deductive or inductive instruction was identified. The data show a clear role for formal linguistic categories in explaining patterns in the data. Implications for fine-tuning instructional intervention and testing of verb classes are discussed.
|
[
"Reasoning",
"Linguistics & Cognitive NLP",
"Linguistic Theories"
] |
[
8,
48,
57
] |
http://arxiv.org/abs/2204.05356v1
|
A Generative Language Model for Few-shot Aspect-Based Sentiment Analysis
|
Sentiment analysis is an important task in natural language processing. In recent works, pre-trained language models are often used to achieve state-of-the-art results, especially when training data is scarce. It is common to fine-tune on the downstream task, usually by adding task-specific layers on top of the model. In this paper, we focus on aspect-based sentiment analysis, which involves extracting aspect term, category, and predicting their corresponding polarities. In particular, we are interested in few-shot settings. We propose to reformulate the extraction and prediction tasks into the sequence generation task, using a generative language model with unidirectional attention (GPT2 is used unless stated otherwise). This way, the model learns to accomplish the tasks via language generation without the need of training task-specific layers. Our evaluation results on the single-task polarity prediction show that our approach outperforms the previous state-of-the-art (based on BERT) on average performance by a large margins in few-shot and full-shot settings. More importantly, our generative approach significantly reduces the model variance caused by low-resource data. We further demonstrate that the proposed generative language model can handle joint and multi-task settings, unlike previous work. We observe that the proposed sequence generation method achieves further improved performances on polarity prediction when the model is trained via joint and multi-task settings. Further evaluation on similar sentiment analysis datasets, SST-2, SST- and OOS intent detection validates the superiority and noise robustness of generative language model in few-shot settings.
|
[
"Language Models",
"Low-Resource NLP",
"Semantic Text Processing",
"Sentiment Analysis",
"Aspect-based Sentiment Analysis",
"Text Generation",
"Responsible & Trustworthy NLP"
] |
[
52,
80,
72,
78,
23,
47,
4
] |
http://arxiv.org/abs/2202.13229v1
|
A Generative Model for Relation Extraction and Classification
|
Relation extraction (RE) is an important information extraction task which provides essential information to many NLP applications such as knowledge base population and question answering. In this paper, we present a novel generative model for relation extraction and classification (which we call GREC), where RE is modeled as a sequence-to-sequence generation task. We explore various encoding representations for the source and target sequences, and design effective schemes that enable GREC to achieve state-of-the-art performance on three benchmark RE datasets. In addition, we introduce negative sampling and decoding scaling techniques which provide a flexible tool to tune the precision and recall performance of the model. Our approach can be extended to extract all relation triples from a sentence in one pass. Although the one-pass approach incurs certain performance loss, it is much more computationally efficient.
|
[
"Relation Extraction",
"Information Retrieval",
"Text Classification",
"Information Extraction & Text Mining"
] |
[
75,
24,
36,
3
] |
SCOPUS_ID:85128324337
|
A Generative Model for Topic Discovery and Polysemy Embeddings on Directed Attributed Networks
|
Combining topic discovery with topic-specific word embeddings is a popular, powerful method for text mining in a small collection of documents. However, the existing researches purely modeled on the contents of documents and led to discovering noisy topics. This paper proposes a generative model, the skip-gram topical word-embedding model (simplified as steoLC) on asymmetric document link networks, where nodes correspond to documents and links refer to directed references between documents. It simultaneously improves the performance of topic discovery and polysemous word embeddings. Each skip-gram in a document is generated based on the topic distribution of the document and the two word embeddings in the skip-gram. Each directed link is generated based on the hidden topic distribution of the beginning document node. For a document, the skip-grams and links share a common topic distribution. Parameter estimation is inferred and an algorithm is designed to learn the model parameters by combining the expectation-maximization (EM) algorithm with the negative sampling method. Experimental results show that our method generates more useful topic-specific word embeddings and coherent latent topics than the state-of-the-art models.
|
[
"Semantic Text Processing",
"Representation Learning"
] |
[
72,
12
] |
SCOPUS_ID:85113495310
|
A Generative Text Summarization Model Based on Document Structure Neural Network
|
Aiming at the low accuracy of the automatic generation of text summaries in the field of data mining, as well as the defects of the existing encoder and decoder models, this paper proposes a generative text summarization model based on the document structure neural network. The model introduces the document structure, divides the text into a word encoding layer and a sentence encoding layer, and builds a top-down hierarchical structure to avoid the back propagation error problem caused by the long input sequence in the traditional encoder and decoder model; At each level, an attention mechanism is added, and a multi-attention mechanism is proposed and introduced, which refines the granularity of the attention mechanism, thereby improving the accuracy of text summary generation. Experimental results show that, compared with the original encoder-decoder model, this model can effectively refines the granularity of the attention mechanism and significantly improve the accuracy of text summary generation.
|
[
"Language Models",
"Semantic Text Processing",
"Summarization",
"Text Generation",
"Information Extraction & Text Mining"
] |
[
52,
72,
30,
47,
3
] |
http://arxiv.org/abs/2210.08692v2
|
A Generative User Simulator with GPT-based Architecture and Goal State Tracking for Reinforced Multi-Domain Dialog Systems
|
Building user simulators (USs) for reinforcement learning (RL) of task-oriented dialog systems (DSs) has gained more and more attention, which, however, still faces several fundamental challenges. First, it is unclear whether we can leverage pretrained language models to design, for example, GPT-2 based USs, to catch up and interact with the recently advanced GPT-2 based DSs. Second, an important ingredient in a US is that the user goal can be effectively incorporated and tracked; but how to flexibly integrate goal state tracking and develop an end-to-end trainable US for multi-domains has remained to be a challenge. In this work, we propose a generative user simulator (GUS) with GPT-2 based architecture and goal state tracking towards addressing the above two challenges. Extensive experiments are conducted on MultiWOZ2.1. Different DSs are trained via RL with GUS, the classic agenda-based user simulator (ABUS) and other ablation simulators respectively, and are compared for cross-model evaluation, corpus-based evaluation and human evaluation. The GUS achieves superior results in all three evaluation tasks.
|
[
"Language Models",
"Natural Language Interfaces",
"Semantic Text Processing",
"Dialogue Systems & Conversational Agents"
] |
[
52,
11,
72,
38
] |
http://arxiv.org/abs/1508.03826v1
|
A Generative Word Embedding Model and its Low Rank Positive Semidefinite Solution
|
Most existing word embedding methods can be categorized into Neural Embedding Models and Matrix Factorization (MF)-based methods. However some models are opaque to probabilistic interpretation, and MF-based methods, typically solved using Singular Value Decomposition (SVD), may incur loss of corpus information. In addition, it is desirable to incorporate global latent factors, such as topics, sentiments or writing styles, into the word embedding model. Since generative models provide a principled way to incorporate latent factors, we propose a generative word embedding model, which is easy to interpret, and can serve as a basis of more sophisticated latent factor models. The model inference reduces to a low rank weighted positive semidefinite approximation problem. Its optimization is approached by eigendecomposition on a submatrix, followed by online blockwise regression, which is scalable and avoids the information loss in SVD. In experiments on 7 common benchmark datasets, our vectors are competitive to word2vec, and better than other MF-based methods.
|
[
"Semantic Text Processing",
"Representation Learning"
] |
[
72,
12
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.