app.py

"""Gradio app that showcases Scandinavian zero-shot text classification models."""

import gradio as gr
from transformers import pipeline
from luga import language as detect_language


# Load the zero-shot classification pipeline
classifier = pipeline(
    "zero-shot-classification", model="alexandrainst/scandi-nli-large"
)


# Set the description for the interface
DESCRIPTION = """Classify text in Danish, Swedish or Norwegian into categories, without
finetuning on any training data!

Note that the models will most likely *not* work as well as a finetuned model on your
specific data, but they can be used as a starting point for your own classification
task ✨


Also, be patient, as this demo is running on a CPU!"""


def classification(task: str, doc: str) -> str:
    """Classify text into categories.

    Args:
        task (str):
            Task to perform.
        doc (str):
            Text to classify.

    Returns:
        str:
            The predicted label.
    """
    # Detect the language of the text
    language = detect_language(doc.replace('\n', ' ')).name

    # Define the confidence string based on the language
    if language == "sv" or language == "no":
        confidence_str = "konfidensnivå"
    else:
        confidence_str = "konfidensniveau"

    # If the task is sentiment, classify the text into positive, negative or neutral
    if task == "Sentiment classification":
        if language == "sv":
            hypothesis_template = "Detta exempel är {}."
            candidate_labels = ["positivt", "negativt", "neutralt"]
        elif language == "no":
            hypothesis_template = "Dette eksemplet er {}."
            candidate_labels = ["positivt", "negativt", "nøytralt"]
        else:
            hypothesis_template = "Dette eksempel er {}."
            candidate_labels = ["positivt", "negativt", "neutralt"]

    # Else if the task is topic, classify the text into a topic
    elif task == "News topic classification":
        if language == "sv":
            hypothesis_template = "Detta exempel handlar om {}."
            candidate_labels = [
                "krig",
                "politik",
                "utbildning",
                "hälsa",
                "ekonomi",
                "mode",
                "sport",
            ]
        elif language == "no":
            hypothesis_template = "Dette eksemplet handler om {}."
            candidate_labels = [
                "krig",
                "politikk",
                "utdanning",
                "helse",
                "økonomi",
                "mote",
                "sport",
            ]
        else:
            hypothesis_template = "Denne nyhedsartikel handler primært om {}."
            candidate_labels = [
                "krig",
                "politik",
                "uddannelse",
                "sundhed",
                "økonomi",
                "mode",
                "sport",
            ]

    # Else if the task is spam detection, classify the text into spam or not spam
    elif task == "Spam detection":
        if language == "sv":
            hypothesis_template = "Det här e-postmeddelandet ser {}."
            candidate_labels = {
                "ut som ett skräppostmeddelande": "Spam",
                "inte ut som ett skräppostmeddelande": "Inte spam",
            }
        elif language == "no":
            hypothesis_template = "Denne e-posten ser {}."
            candidate_labels = {
                "ut som en spam-e-post": "Spam",
                "ikke ut som en spam-e-post": "Ikke spam",
            }
        else:
            hypothesis_template = "Denne e-mail ligner {}."
            candidate_labels = {
                "en spam e-mail": "Spam",
                "ikke en spam e-mail": "Ikke spam",
            }

    # Else if the task is product feedback detection, classify the text into product
    # feedback or not product feedback
    elif task == "Product feedback detection":
        if language == "sv":
            hypothesis_template = "Den här kommentaren är {}."
            candidate_labels = {
                "en recension av en produkt": "Produktfeedback",
                "inte en recension av en produkt": "Inte produktfeedback",
            }
        elif language == "no":
            hypothesis_template = "Denne kommentaren er {}."
            candidate_labels = {
                "en anmeldelse av et produkt": "Produkttilbakemelding",
                "ikke en anmeldelse av et produkt": "Ikke produkttilbakemelding",
            }
        else:
            hypothesis_template = "Denne kommentar er {}."
            candidate_labels = {
                "en anmeldelse af et produkt": "Produktfeedback",
                "ikke en anmeldelse af et produkt": "Ikke produktfeedback",
            }

    # Else the task is not supported, so raise an error
    else:
        raise ValueError(f"Task {task} not supported.")

    # If `candidate_labels` is a list then convert it to a dictionary, where the keys
    # are the entries in the list and the values are the keys capitalized
    if isinstance(candidate_labels, list):
        candidate_labels = {label: label.capitalize() for label in candidate_labels}

    # Run the classifier on the text
    result = classifier(
        doc,
        candidate_labels=list(candidate_labels.keys()),
        hypothesis_template=hypothesis_template,
    )

    print(result)

    # Return the predicted label
    return (
        f"{candidate_labels[result['labels'][0]]}\n"
        f"({confidence_str}: {result['scores'][0]:.0%})"
    )

# Create a dropdown menu for the task
dropdown = gr.inputs.Dropdown(
    label="Task",
    choices=[
        "Sentiment classification",
        "News topic classification",
        "Spam detection",
        "Product feedback detection",
    ],
    default="Sentiment classification",
)

# Create a text box for the input text
input_textbox = gr.inputs.Textbox(
    label="Text", default="Jeg er helt vild med fodbolden 😊"
)

# Create the interface, where the function depends on the task chosen
interface = gr.Interface(
    fn=classification,
    inputs=[dropdown, input_textbox],
    outputs=gr.outputs.Label(type="text"),
    title="Scandinavian zero-shot text classification",
    description=DESCRIPTION,
)

# Run the app
interface.launch()