{ "cells": [ { "cell_type": "markdown", "id": "a5c7b436", "metadata": {}, "source": [ "# Analysis of Retrieval Results" ] }, { "cell_type": "markdown", "id": "6082ebb6", "metadata": {}, "source": [ "## Load annotations and retrieval results" ] }, { "cell_type": "code", "execution_count": 1, "id": "fefb786d", "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
UUIDFILEQUOTE_TRANSCRIPTIONANCHORCOMMENTSTAGSPAGETRANSCRIPTION_CLEANED
5512faf6dce-7806-48b6-8e8b-2bf926fc88fcZ166069305_merged.txtNegypten aber hat nicht allein Wunder der Kunſ...char-offset:2567p. 8'plants vegetation 25G'8('n? Negypten aber hat nicht allein Wunder der...
5522e026f81-cbe7-43ba-9b37-5f5bd2760119Z166069305_merged.txtBarbaren bewohnen jezt das Land und Räuber beh...char-offset:3391p. 8'plants vegetation 25G'8('r. Barbaren bewohnen jezt das Land und Raube...
5530d5708db-1bf6-4de4-ad64-9759ebb4fa3cZ166069305_merged.txtDie Waffen eines Mamefu den zu Pferde find zwe...char-offset:5538p. 9-10\\n'animals 25F', 'horses and kindred animals 46C...10('n. Die Waffen eines Mamefu den zu Pferde fin...
55464b5828e-b14d-49dc-8d88-f97de52f9202Z166069305_merged.txtDen Zaum reines Pferdes nimmt er zwiſchen die ...char-offset:5987p. 10\\n'animals 25F', 'horses and kindred animals 46C...10('l. Den Zaum reines Pferdes nimmt er zwischen...
555fa2e8b0d-d8aa-46c8-ac97-1a8e2b14600dZ166069305_merged.txtSeine Häuſer ſind arms felige Kothhütten, und ...char-offset:8769p. 11-12'plants vegetation 25G'12(' Hauser sind arms felige Kothhutten, und fei...
...........................
11173d4a21ad-ddd8-4d3a-b018-d7dbb9f186dcZ166069305_merged.txtBittt adit2: 001), ont ma Des Abends begaben ſ...char-offset:821475NaN'animals 25F', 'birds 25F3'504(\"00 Bittt adit2: 001), ont ma Des Abends bega...
11182aefbba4-e796-4add-bdb1-0768afe0a3ceZ166069305_merged.txt57 3 Chazelle lieb Chazelleschar-offset:823734NaN'animals 25F', 'hoofed animalGAZELLE 25F24'504('ue 57 3 Chazelle lieb Chazelles 65', 814073)
1119f50b57c6-d751-4d41-8670-8b5254d964ecZ166069305_merged.txtEin' to svil, dulu UPUN Fahn irlichin( jiyi vo...char-offset:824751NaN'animals 25F'504(\"c. Ein' to suil, dulu UPUN Fahn irlichin( ji...
1120d8cfcc30-5f00-4681-826b-c5c7a87e768bZ166069305_merged.txtDez HENNÈI, Blüthen. Eine Blüthe ron natiiitic...char-offset:824901NaN'plants vegetation 25G'504('-- Dez HENNEI, Bluthen. Eine Bluthe ron nati...
1121bbf95b5f-40d1-476a-a392-2dabb1278c16Z166069305_merged.txtRiedel fee. Fig. 1, ATLÉ, A, Stamn.2.7weig. Fi...char-offset:825427NaN'plants vegetation 25G'504(\" 1 Riedel fee. Fig. 1, ATLE, A, Stamn.2.7wei...
\n", "

571 rows × 8 columns

\n", "
" ], "text/plain": [ " UUID FILE \\\n", "551 2faf6dce-7806-48b6-8e8b-2bf926fc88fc Z166069305_merged.txt \n", "552 2e026f81-cbe7-43ba-9b37-5f5bd2760119 Z166069305_merged.txt \n", "553 0d5708db-1bf6-4de4-ad64-9759ebb4fa3c Z166069305_merged.txt \n", "554 64b5828e-b14d-49dc-8d88-f97de52f9202 Z166069305_merged.txt \n", "555 fa2e8b0d-d8aa-46c8-ac97-1a8e2b14600d Z166069305_merged.txt \n", "... ... ... \n", "1117 3d4a21ad-ddd8-4d3a-b018-d7dbb9f186dc Z166069305_merged.txt \n", "1118 2aefbba4-e796-4add-bdb1-0768afe0a3ce Z166069305_merged.txt \n", "1119 f50b57c6-d751-4d41-8670-8b5254d964ec Z166069305_merged.txt \n", "1120 d8cfcc30-5f00-4681-826b-c5c7a87e768b Z166069305_merged.txt \n", "1121 bbf95b5f-40d1-476a-a392-2dabb1278c16 Z166069305_merged.txt \n", "\n", " QUOTE_TRANSCRIPTION ANCHOR \\\n", "551 Negypten aber hat nicht allein Wunder der Kunſ... char-offset:2567 \n", "552 Barbaren bewohnen jezt das Land und Räuber beh... char-offset:3391 \n", "553 Die Waffen eines Mamefu den zu Pferde find zwe... char-offset:5538 \n", "554 Den Zaum reines Pferdes nimmt er zwiſchen die ... char-offset:5987 \n", "555 Seine Häuſer ſind arms felige Kothhütten, und ... char-offset:8769 \n", "... ... ... \n", "1117 Bittt adit2: 001), ont ma Des Abends begaben ſ... char-offset:821475 \n", "1118 57 3 Chazelle lieb Chazelles char-offset:823734 \n", "1119 Ein' to svil, dulu UPUN Fahn irlichin( jiyi vo... char-offset:824751 \n", "1120 Dez HENNÈI, Blüthen. Eine Blüthe ron natiiitic... char-offset:824901 \n", "1121 Riedel fee. Fig. 1, ATLÉ, A, Stamn.2.7weig. Fi... char-offset:825427 \n", "\n", " COMMENTS TAGS PAGE \\\n", "551 p. 8 'plants vegetation 25G' 8 \n", "552 p. 8 'plants vegetation 25G' 8 \n", "553 p. 9-10\\n 'animals 25F', 'horses and kindred animals 46C... 10 \n", "554 p. 10\\n 'animals 25F', 'horses and kindred animals 46C... 10 \n", "555 p. 11-12 'plants vegetation 25G' 12 \n", "... ... ... ... \n", "1117 NaN 'animals 25F', 'birds 25F3' 504 \n", "1118 NaN 'animals 25F', 'hoofed animalGAZELLE 25F24' 504 \n", "1119 NaN 'animals 25F' 504 \n", "1120 NaN 'plants vegetation 25G' 504 \n", "1121 NaN 'plants vegetation 25G' 504 \n", "\n", " TRANSCRIPTION_CLEANED \n", "551 ('n? Negypten aber hat nicht allein Wunder der... \n", "552 ('r. Barbaren bewohnen jezt das Land und Raube... \n", "553 ('n. Die Waffen eines Mamefu den zu Pferde fin... \n", "554 ('l. Den Zaum reines Pferdes nimmt er zwischen... \n", "555 (' Hauser sind arms felige Kothhutten, und fei... \n", "... ... \n", "1117 (\"00 Bittt adit2: 001), ont ma Des Abends bega... \n", "1118 ('ue 57 3 Chazelle lieb Chazelles 65', 814073) \n", "1119 (\"c. Ein' to suil, dulu UPUN Fahn irlichin( ji... \n", "1120 ('-- Dez HENNEI, Bluthen. Eine Bluthe ron nati... \n", "1121 (\" 1 Riedel fee. Fig. 1, ATLE, A, Stamn.2.7wei... \n", "\n", "[571 rows x 8 columns]" ] }, "execution_count": 1, "metadata": {}, "output_type": "execute_result" } ], "source": [ "import pandas as pd\n", "import os\n", "import re\n", "\n", "text_annotations_path = \"data/annotations/\"\n", "filename = 'DHd2025_referenceReports_merged-cleaned__annotations.xlsx'\n", "\n", "annotations = pd.read_excel(os.path.join(text_annotations_path, filename))\n", "\n", "## Remove unwanted tags from list\n", "unwanted_strings = ['pages', 'subheading', 'footnote', 'table of contents', 'footnote irrelevant', 'title', 'scan mistake', 'dictionary']\n", "\n", "# Remove rows where 'TAGS' contains any unwanted string\n", "annotations = annotations[~annotations['TAGS'].str.contains('|'.join(unwanted_strings), case=False, na=False)].reset_index(drop=True)\n", "annotations = annotations[annotations['FILE'].str.contains('Z166069305', case=False, na=False)]\n", "\n", "annotations" ] }, { "cell_type": "code", "execution_count": 2, "id": "c6bff210", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Number of excluded sentences: 0\n", "Number of annotated mentions: 48\n" ] }, { "data": { "text/plain": [ "{'id': '0d5708db-1bf6-4de4-ad64-9759ebb4fa3c',\n", " 'annotated_text': \"('n. Die Waffen eines Mamefu den zu Pferde find zwei --- 00010_page10_cleaned.txt --- grose Flinten, die ihm seine Diener zur Seite nach: tragen und die er ein einziges mal losschiest. Da', 5402)\",\n", " 'comment': 'p. 9-10\\n',\n", " 'page': 10,\n", " 'file': 'Z166069305_merged.txt'}" ] }, "execution_count": 2, "metadata": {}, "output_type": "execute_result" } ], "source": [ "## Filter the DataFrame\n", "filtered_df = annotations[annotations['TAGS'].str.contains('horse', na=False)]\n", "# Count the number of rows before filtering short sentences\n", "before_filter_count = len(filtered_df)\n", "\n", "# Filter sentences that are 5 words or longer in 'QUOTE_TRANSCRIPTION'\n", "filtered_df = filtered_df[filtered_df['TRANSCRIPTION_CLEANED'].apply(lambda x: len(x.split()) >= 5)]\n", "# Count the number of rows after filtering short sentences\n", "after_filter_count = len(filtered_df)\n", "excluded_count = before_filter_count - after_filter_count\n", "print(f\"Number of excluded sentences: {excluded_count}\")\n", "\n", "mentions_dict = filtered_df.set_index('UUID').to_dict(orient='index')\n", "\n", "mentions = [{\"id\": key, \"annotated_text\": v['TRANSCRIPTION_CLEANED'], \"comment\": v['COMMENTS'], \"page\": v['PAGE'], \"file\": v['FILE']} for key, v in mentions_dict.items()]\n", "\n", "print(f\"Number of annotated mentions: {len(filtered_df)}\")\n", "mentions[0]" ] }, { "cell_type": "code", "execution_count": 3, "id": "cc29b140", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Total annotations: 48\n", "On total pages: 40\n" ] } ], "source": [ "## Summary of annotations\n", "\n", "# Total annotations:\n", "# Remove duplicates based on \"page\"\n", "unique_annot = {entry[\"id\"]: entry for entry in mentions}.values()\n", "\n", "# Remove duplicates based on \"page\"\n", "unique_pages_annot = {entry[\"page\"]: entry for entry in mentions}.values()\n", "\n", "# Extract the \"page\" value from each dictionary of unique pages found\n", "unique_pages_list = [item['page'] for item in unique_pages_annot]\n", "\n", "print(f'Total annotations: {len(unique_annot)}')\n", "print(f'On total pages: {len(unique_pages_annot)}')" ] }, { "cell_type": "code", "execution_count": 5, "id": "bd863c3b", "metadata": {}, "outputs": [], "source": [ "## Load retrieval results with Marqo DHd2025\n", "\n", "## Pferd, Pferde\n", "retr_orig = pd.read_csv('data/retrieval_results/sonnini_original_OCR/i_onit-test-index-sonnini-q_Pferd-Pferde.csv')\n", "retr_prep = pd.read_csv('data/retrieval_results/sonnini_llm_corrected/i_onit-sonnini-DHd2025-prep-q_Pferd, Pferde.csv')\n", "retr_clean = pd.read_csv('data/retrieval_results/sonnini_cleaned/i_onit-sonnini-DHd2025-clean-q_Pferd, Pferde.csv')" ] }, { "cell_type": "markdown", "id": "6b4a54bd", "metadata": {}, "source": [ "## Evaluation of results" ] }, { "cell_type": "code", "execution_count": 7, "id": "296250aa", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "200\n" ] }, { "data": { "text/plain": [ "{'id': '01b71d37-9ead-4eb2-8504-d16cbd00e866',\n", " 'text_document': 'befand mich in einiger Entfernung davon, und der Ueber- rest unserer Begleitung folgte in ziemlicher Weite nach. Ein Trupp Beduinen zu Pferde. brach auf einmal hinter den Mauern hervor. Ich konnte sie an. fanglich unter den Staubwolken, die sie erregten, nicht unterscheiden: als fie fich aber ausgebreitet hatten, erfannte ich sowohl die Ort als die Anzahl von Leuten, mit denen wir zu thun haben sollten. Ich wandte sogleich inein Pferd um, und mit diefem vortreflichen Reuter, auf welchem man mich nicht hatte einbohlen konnen, hatte ich bald meine Gefahrten erreicht, die auch von ihren Came len herab diese. Reuterei gewahr worden waren. Ich fand sie zu Fus und in geschlossene Glieder gestellt. Ich sturzte mich von meinem Pferde herab und munterte fie zu einenu muthigen Widerstande auf. Wir waren im Ganzen rechs Personen, unter welchen wir nur auf drei | rechnen konnten. Zwei Eingebohrne fonnten nicht viel helfen, und der Zeichner war noch jung und ohne Erfah- rung, und konnte keine Flinte losschiesen. ut Die Standhaftigkeit einer Handvou Menschen, die fich mitten in einer Sandflache allein befanden, und die von keiner Seite gedekt waren, liberraschte eine Schwa: dron von ungefahr hundert Beduinen: man kann daraus schliesen, wie wenig furchtbar solche Feinde find, deren Muth fich blos darauf beschrankt, das fie fich in zahlreichen Horden versammeln, um eine leichte Plundes rung zu veruben, und auf eine feige Art den kleinen Krieg zu fuhren. Ob fie gleich in starkem Galop auf uns zuge. ritten kamen, so machten sie doch plozlich, etwann hundert Schritte von uns, halt. Sie riefen uns zu, kein Feuer zu geben, und ich erwiederte ihnen, fie sollten nicht vors rucken. Sie blieben einige Augenblicke in einer Art von Ungewisheit, was sie thun souten, wahrend welcher wir fie unter einander berathschlagen sahen. Sie theilten sich',\n", " 'text_vector': 'befand mich in einiger Entfernung davon, und der Ueber- rest unserer Begleitung folgte in ziemlicher Weite nach. Ein Trupp Beduinen zu Pferde.',\n", " 'page': 430,\n", " 'rerank': 1,\n", " 'barcode': 'Z166069305',\n", " 'onb_link': 'https://digital.onb.ac.at/OnbViewer/viewer.faces?doc=ABO_%2BZ166069305'}" ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" } ], "source": [ "## Analyse overlap of retrieved texts with annotated texts\n", "\n", "\"\"\"\n", "In this notebook, we attempted to compare the heterogeneous texts that were annotated during the project\n", "with the retrieval results by comparing n-word phrases from the annotated sentences with the retrieved\n", "sentence vectors. This approach is experimental. An alternative evaluation by calculating the Levenshtein\n", "distance between the annotated text vectors and the retrieved text vectors will be done in future.\n", "\"\"\"\n", "\n", "## Create dict from retrieved texts\n", "#retr = retr_orig.head(200)\n", "#retr = retr_prep.head(200)\n", "retr = retr_clean.head(200)\n", "\n", "retrieved_dict = retr.set_index('_id').to_dict(orient='index')\n", "\n", "retrieved = [\n", " {\n", " \"id\": key,\n", " \"text_document\": v['text_clean'], # 'text', 'text_clean'\n", " \"text_vector\": v['unpacked_highlights'] if pd.notna(v['unpacked_highlights']) else v['text'],\n", " \"page\": v['page'], #int((re.search(r'\\d+', v['page'])).group()),\n", " \"rerank\": v['rerank'],\n", " \"barcode\": v['barcode'],\n", " \"onb_link\": v['onb_viewer_link']\n", " }\n", " for key, v in retrieved_dict.items()\n", "]\n", "\n", "print(len(retrieved))\n", "retrieved[0]" ] }, { "cell_type": "code", "execution_count": 8, "id": "307ab5c6", "metadata": { "code_folding": [] }, "outputs": [], "source": [ "## Function to generate n-word phrases from a sentence\n", "def generate_n_word_phrases(sentence, n):\n", " words = sentence.lower().split()\n", " return [' '.join(words[i:i+n]) for i in range(len(words) - n + 1)]\n", "\n", "## Function to check if any of the n-word phrases is in the text_vector\n", "## and store the first matching phrase\n", "def check_sentence_in_text_vector(sentence, text_vector, n):\n", " text_vector_lower = text_vector.lower()\n", " phrases = generate_n_word_phrases(sentence, n)\n", " for phrase in phrases:\n", " if phrase in text_vector_lower:\n", " return True, phrase # Return True and the matching phrase\n", " return False, None # No match found, return None\n", " #return any(phrase in text_vector_lower for phrase in phrases)\n", "\n", "# Number of words in the phrase to check\n", "n = 5 # Change this to the desired number of words\n", "\n", "# List to store results\n", "results = []\n", "\n", "# Iterate through each sentence in the mentions list\n", "for sentence in mentions:\n", " found = False\n", " # Check if any n-word phrase from this sentence is present in any of the text_vector fields\n", " for entry in retrieved:\n", " #print(entry['text_vector'])\n", " match_found, match = check_sentence_in_text_vector(sentence['annotated_text'], entry['text_vector'], n)\n", " if match_found == True:\n", " found = True\n", " break # No need to check further if a match is found\n", " if found == True:\n", " results.append({\n", " \"sentence_annotation\": sentence['annotated_text'],\n", " \"id_annotation\": sentence['id'],\n", " \"page_annotation\": sentence['page'],\n", " \"comment\": sentence['comment'],\n", " \"found_in_text_vector\": found,\n", " \"text_vector\": entry['text_vector'],\n", " \"matching_phrase\": match,\n", " \"id_text_vector\": entry['id'],\n", " \"barcode\": entry['barcode'],\n", " \"page_text\": entry['page'],\n", " \"onb_link\": entry['onb_link'],\n", " \"file_annotation\": sentence['file']\n", " })\n", " else:\n", " results.append({\n", " \"sentence_annotation\": sentence['annotated_text'],\n", " \"id_annotation\": sentence['id'],\n", " \"page_annotation\": sentence['page'],\n", " \"file_annotation\": sentence['file'],\n", " \"found_in_text_vector\": found\n", " })\n", "\n", "retrieved_df = pd.DataFrame(retrieved)\n", "retrieved_df.rename(columns={'id': 'id_text_vector'}, inplace=True)\n", "mat_df = pd.DataFrame(results)\n", "matches_df = retrieved_df.merge(mat_df, on='id_text_vector', how='outer')\n", "\n", "# Sort the DataFrame based on the 'rrf_rank' column in ascending order\n", "matches_df = matches_df.sort_values(by='rerank', ascending=True)" ] }, { "cell_type": "code", "execution_count": 9, "id": "68868787", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "True Positives: Found 27 relevant mentions on 27 pages of 48 relevant mentions on 40 pages\n", "(67.50%).\n", "Found average of 1.00 found mentions per page vs. GT of 1.20 annotations per page.\n" ] }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
sentence_annotationtext_vector_xrerankid_annotationpagepage_annotationfile_annotation
0('h. Ein Trupp Beduinen zu Pferde. brach auf e...befand mich in einiger Entfernung davon, und d...1.0ffbabe61-0a21-4e4f-b6e4-2ddc4724b40c430.0434.0Z166069305_merged.txt
1('n. Man muntert beide auch mit einem Klatsche...Man muntert beide auch mit einem Klatschen der...2.09a41db07-a5c8-4b3f-931a-43418f800768399.0403.0Z166069305_merged.txt
2('t. Ein kriegerisches und eroberungssuchtiges...Eine weife Nation hingegen, die den Ackerbau f...3.02caacee3-c22d-4a76-a2d7-ba794856a103175.0176.0Z166069305_merged.txt
3('n. Die Waffen eines Mamefu den zu Pferde fin...Die Waffen eines Mamefu den zu Pferde find zwei4.00d5708db-1bf6-4de4-ad64-9759ebb4fa3c9.010.0Z166069305_merged.txt
4('r. Das heise Clima dieser Gegenden des Alter...Das Pferd, das eben fo feurig ist als die Luft...5.04f7c0172-b25f-4e62-9cb5-a1c1213c6d7d220.0222.0Z166069305_merged.txt
5(\"n. Diese Banditen waren nicht damit zufriede...Andere Uraber erwiesen meinen Gefahrten diesel...6.0c86544f5-33f0-4d9a-b048-4995ae36c493435.0439.0Z166069305_merged.txt
6('s. ?. Wir fahen zwei Beduinen zu Pferde; fie...?. Wir fahen zwei Beduinen zu Pferde; fie floh...7.03a1cf968-2b82-48d9-a784-d3a7f5198890413.0417.0Z166069305_merged.txt
7(\"b. Der Hraber muste mir fur den taglichen Pr...Der Hraber muste mir fur den taglichen Preis v...8.0b0450149-d074-4079-87c4-cdc252ce7c67383.0387.0Z166069305_merged.txt
8('r. Ich war mit meinem Pferde uorausgeritten,...Ich war mit meinem Pferde vorausgeritten, und ...9.0a16c02e9-16a6-4780-b65e-51510b8db6fb406.0410.0Z166069305_merged.txt
9('-- Das Lager stand etwan einige hundert Fus ...Man bauete verschiedente Arten Viehfutter dara...10.0b8a8f0ea-926e-4d89-b21d-e89e69a35c2f394.0397.0Z166069305_merged.txt
10(\"n. Ich bemerkte oft frische Spuren uon Strau...Ihre Jagd ist eine von den Leibesubuns gen, wo...11.02703eecd-2bb6-432b-beaf-c92933736058424.0428.0Z166069305_merged.txt
11('b. Ich bestieg ein sehr schones Pferd, und i...Ich bestieg ein sehr schones Pferd, und ich wu...12.03f08cff7-7220-4c5e-9d1f-1b8d4fc92028405.0409.0Z166069305_merged.txt
12('n. Sie halten sich mit ihren trefa lich klei...Sie halten sich mit ihren trefa lich kleinen P...13.028370a93-4521-48d0-a8da-50545822debc12.012.0Z166069305_merged.txt
13('n. Es war eine sehr gunstige Gelegenheit: di...mittel herum, die man ihnen gereicht hatte: ih...14.0d9613671-efc9-4a0e-9770-d2b8432d71de455.0460.0Z166069305_merged.txt
14('r. Ihre ganze Nahrung besteht in Brodt oder ...Und wirklich sind diejenigen, die in der Wuste...15.05913d181-6da7-4169-9187-9eb885d33e33448.0452.0Z166069305_merged.txt
15('g. Sie waren bald bei uns und hatten uns, oh...Es verdros mich, das ich mich solchen Raubern ...16.01a1ef27f-0449-4664-8f6c-e929f089504a431.0435.0Z166069305_merged.txt
17('d. Der Efel, der den Munduorrath trug und ei...Der Efel, der den Mundvorrath trug und einer l...18.05892d1ea-8016-4e3e-9176-f6f6df833125148.0149.0Z166069305_merged.txt
18('t. Man reitet zu Pferde durch dieselbe, wenn...Man reitet zu Pferde durch dieselbe, wenn nich...19.0e422911b-8c7b-4a1e-a4aa-df3d3f617c51152.0153.0Z166069305_merged.txt
19(\"n. Eine Viertelmeile weit richteten wir unse...Eine Viertelmeile weit richteten wir unsern We...20.02602d32c-2c40-4add-893c-abfc115b3c38415.0419.0Z166069305_merged.txt
20(\"n. Auf den Strasen uerkaufte man Bokshornste...Auf den Strasen verkaufte man Bokshornstengel*...21.0b3a06b70-a367-4eef-a21e-788142390022261.0263.0Z166069305_merged.txt
21('l. Den Zaum reines Pferdes nimmt er zwischen...Sein letztes Hulfeitrittet find jwei Sa: bel. ...22.064b5828e-b14d-49dc-8d88-f97de52f920210.010.0Z166069305_merged.txt
22('n. Es giebt ein gutes Futter, das die Pferde...machten Eigenschaften verdient der Anbau das B...23.03bf200db-aeb9-4823-8923-6458b1f01d7e263.0266.0Z166069305_merged.txt
24('-- Wir kamen aus einer Stadt, wo die Menge u...Wir kamen aus einer Stadt, wo die Menge von Ma...25.0cdd8f310-f15e-44d7-bd06-3e91351e3eae59.059.0Z166069305_merged.txt
25(\"n. Den Sag nach der Ankunft dieses Mannes en...Den Sag nach der Ankunft dieses Mannes entdekt...26.059c9905e-6cfa-4544-8805-dfbd53e7ed29454.0459.0Z166069305_merged.txt
26('n. Die Araber uerfolgen fte zu Pferde; und m...Sie entfernen Fich aber sogleich, fo bald fie ...27.063f2c3e0-69a8-4a22-a356-7d0dd570dee1422.0426.0Z166069305_merged.txt
33(\"t. Jene artige Art uon uierfusigent Thieren,...brauch von ihnen zu machen. Diese Fufe' find w...34.01f056aa1-5980-4f62-9bff-147a9b409fd9423.0427.0Z166069305_merged.txt
62('t. Wir erkletterten auf Maulefeln die Sandhu...Wir erkletterten auf Maulefeln die Sandhugel, ...63.0e4074c9a-c71c-4076-a22e-dce3f7e263c9264.0267.0Z166069305_merged.txt
\n", "
" ], "text/plain": [ " sentence_annotation \\\n", "0 ('h. Ein Trupp Beduinen zu Pferde. brach auf e... \n", "1 ('n. Man muntert beide auch mit einem Klatsche... \n", "2 ('t. Ein kriegerisches und eroberungssuchtiges... \n", "3 ('n. Die Waffen eines Mamefu den zu Pferde fin... \n", "4 ('r. Das heise Clima dieser Gegenden des Alter... \n", "5 (\"n. Diese Banditen waren nicht damit zufriede... \n", "6 ('s. ?. Wir fahen zwei Beduinen zu Pferde; fie... \n", "7 (\"b. Der Hraber muste mir fur den taglichen Pr... \n", "8 ('r. Ich war mit meinem Pferde uorausgeritten,... \n", "9 ('-- Das Lager stand etwan einige hundert Fus ... \n", "10 (\"n. Ich bemerkte oft frische Spuren uon Strau... \n", "11 ('b. Ich bestieg ein sehr schones Pferd, und i... \n", "12 ('n. Sie halten sich mit ihren trefa lich klei... \n", "13 ('n. Es war eine sehr gunstige Gelegenheit: di... \n", "14 ('r. Ihre ganze Nahrung besteht in Brodt oder ... \n", "15 ('g. Sie waren bald bei uns und hatten uns, oh... \n", "17 ('d. Der Efel, der den Munduorrath trug und ei... \n", "18 ('t. Man reitet zu Pferde durch dieselbe, wenn... \n", "19 (\"n. Eine Viertelmeile weit richteten wir unse... \n", "20 (\"n. Auf den Strasen uerkaufte man Bokshornste... \n", "21 ('l. Den Zaum reines Pferdes nimmt er zwischen... \n", "22 ('n. Es giebt ein gutes Futter, das die Pferde... \n", "24 ('-- Wir kamen aus einer Stadt, wo die Menge u... \n", "25 (\"n. Den Sag nach der Ankunft dieses Mannes en... \n", "26 ('n. Die Araber uerfolgen fte zu Pferde; und m... \n", "33 (\"t. Jene artige Art uon uierfusigent Thieren,... \n", "62 ('t. Wir erkletterten auf Maulefeln die Sandhu... \n", "\n", " text_vector_x rerank \\\n", "0 befand mich in einiger Entfernung davon, und d... 1.0 \n", "1 Man muntert beide auch mit einem Klatschen der... 2.0 \n", "2 Eine weife Nation hingegen, die den Ackerbau f... 3.0 \n", "3 Die Waffen eines Mamefu den zu Pferde find zwei 4.0 \n", "4 Das Pferd, das eben fo feurig ist als die Luft... 5.0 \n", "5 Andere Uraber erwiesen meinen Gefahrten diesel... 6.0 \n", "6 ?. Wir fahen zwei Beduinen zu Pferde; fie floh... 7.0 \n", "7 Der Hraber muste mir fur den taglichen Preis v... 8.0 \n", "8 Ich war mit meinem Pferde vorausgeritten, und ... 9.0 \n", "9 Man bauete verschiedente Arten Viehfutter dara... 10.0 \n", "10 Ihre Jagd ist eine von den Leibesubuns gen, wo... 11.0 \n", "11 Ich bestieg ein sehr schones Pferd, und ich wu... 12.0 \n", "12 Sie halten sich mit ihren trefa lich kleinen P... 13.0 \n", "13 mittel herum, die man ihnen gereicht hatte: ih... 14.0 \n", "14 Und wirklich sind diejenigen, die in der Wuste... 15.0 \n", "15 Es verdros mich, das ich mich solchen Raubern ... 16.0 \n", "17 Der Efel, der den Mundvorrath trug und einer l... 18.0 \n", "18 Man reitet zu Pferde durch dieselbe, wenn nich... 19.0 \n", "19 Eine Viertelmeile weit richteten wir unsern We... 20.0 \n", "20 Auf den Strasen verkaufte man Bokshornstengel*... 21.0 \n", "21 Sein letztes Hulfeitrittet find jwei Sa: bel. ... 22.0 \n", "22 machten Eigenschaften verdient der Anbau das B... 23.0 \n", "24 Wir kamen aus einer Stadt, wo die Menge von Ma... 25.0 \n", "25 Den Sag nach der Ankunft dieses Mannes entdekt... 26.0 \n", "26 Sie entfernen Fich aber sogleich, fo bald fie ... 27.0 \n", "33 brauch von ihnen zu machen. Diese Fufe' find w... 34.0 \n", "62 Wir erkletterten auf Maulefeln die Sandhugel, ... 63.0 \n", "\n", " id_annotation page page_annotation \\\n", "0 ffbabe61-0a21-4e4f-b6e4-2ddc4724b40c 430.0 434.0 \n", "1 9a41db07-a5c8-4b3f-931a-43418f800768 399.0 403.0 \n", "2 2caacee3-c22d-4a76-a2d7-ba794856a103 175.0 176.0 \n", "3 0d5708db-1bf6-4de4-ad64-9759ebb4fa3c 9.0 10.0 \n", "4 4f7c0172-b25f-4e62-9cb5-a1c1213c6d7d 220.0 222.0 \n", "5 c86544f5-33f0-4d9a-b048-4995ae36c493 435.0 439.0 \n", "6 3a1cf968-2b82-48d9-a784-d3a7f5198890 413.0 417.0 \n", "7 b0450149-d074-4079-87c4-cdc252ce7c67 383.0 387.0 \n", "8 a16c02e9-16a6-4780-b65e-51510b8db6fb 406.0 410.0 \n", "9 b8a8f0ea-926e-4d89-b21d-e89e69a35c2f 394.0 397.0 \n", "10 2703eecd-2bb6-432b-beaf-c92933736058 424.0 428.0 \n", "11 3f08cff7-7220-4c5e-9d1f-1b8d4fc92028 405.0 409.0 \n", "12 28370a93-4521-48d0-a8da-50545822debc 12.0 12.0 \n", "13 d9613671-efc9-4a0e-9770-d2b8432d71de 455.0 460.0 \n", "14 5913d181-6da7-4169-9187-9eb885d33e33 448.0 452.0 \n", "15 1a1ef27f-0449-4664-8f6c-e929f089504a 431.0 435.0 \n", "17 5892d1ea-8016-4e3e-9176-f6f6df833125 148.0 149.0 \n", "18 e422911b-8c7b-4a1e-a4aa-df3d3f617c51 152.0 153.0 \n", "19 2602d32c-2c40-4add-893c-abfc115b3c38 415.0 419.0 \n", "20 b3a06b70-a367-4eef-a21e-788142390022 261.0 263.0 \n", "21 64b5828e-b14d-49dc-8d88-f97de52f9202 10.0 10.0 \n", "22 3bf200db-aeb9-4823-8923-6458b1f01d7e 263.0 266.0 \n", "24 cdd8f310-f15e-44d7-bd06-3e91351e3eae 59.0 59.0 \n", "25 59c9905e-6cfa-4544-8805-dfbd53e7ed29 454.0 459.0 \n", "26 63f2c3e0-69a8-4a22-a356-7d0dd570dee1 422.0 426.0 \n", "33 1f056aa1-5980-4f62-9bff-147a9b409fd9 423.0 427.0 \n", "62 e4074c9a-c71c-4076-a22e-dce3f7e263c9 264.0 267.0 \n", "\n", " file_annotation \n", "0 Z166069305_merged.txt \n", "1 Z166069305_merged.txt \n", "2 Z166069305_merged.txt \n", "3 Z166069305_merged.txt \n", "4 Z166069305_merged.txt \n", "5 Z166069305_merged.txt \n", "6 Z166069305_merged.txt \n", "7 Z166069305_merged.txt \n", "8 Z166069305_merged.txt \n", "9 Z166069305_merged.txt \n", "10 Z166069305_merged.txt \n", "11 Z166069305_merged.txt \n", "12 Z166069305_merged.txt \n", "13 Z166069305_merged.txt \n", "14 Z166069305_merged.txt \n", "15 Z166069305_merged.txt \n", "17 Z166069305_merged.txt \n", "18 Z166069305_merged.txt \n", "19 Z166069305_merged.txt \n", "20 Z166069305_merged.txt \n", "21 Z166069305_merged.txt \n", "22 Z166069305_merged.txt \n", "24 Z166069305_merged.txt \n", "25 Z166069305_merged.txt \n", "26 Z166069305_merged.txt \n", "33 Z166069305_merged.txt \n", "62 Z166069305_merged.txt " ] }, "execution_count": 9, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Filter rows where 'found_in_text_vector' is True --> True Positives\n", "found = matches_df[matches_df['found_in_text_vector'] == True]\n", "found_fil = found['id_annotation'].drop_duplicates(keep='first')\n", "found_pages = found['page'].drop_duplicates(keep='first')\n", "\n", "# Extract the \"page\" value from each dictionary of unique pages found\n", "unique_foundPages_list = found_pages.to_list()\n", "\n", "# Filter rows where the absolute difference between 'page_annotation' and 'page_text' is less than or equal to 4\n", "#found = found[abs(found['page_annotation'] - found['page_text']) <= 6]\n", "\n", "print(f\"True Positives: Found {len(found_fil)} relevant mentions on {len(unique_foundPages_list)} pages of {len(unique_annot)} relevant mentions on {len(unique_pages_list)} pages\")\n", "print(f\"({len(unique_foundPages_list)/len(unique_pages_list)*100:.2f}%).\")\n", "print(f\"Found average of {len(found_fil)/len(found_pages):.2f} found mentions per page vs. GT of {len(unique_annot)/len(unique_pages_annot):.2f} annotations per page.\")\n", "\n", "found[['sentence_annotation', 'text_vector_x', 'rerank', 'id_annotation', 'page', 'page_annotation', 'file_annotation']]" ] }, { "cell_type": "code", "execution_count": 10, "id": "1e9beb8e", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "False Negatives: Missed 21 mentions on 20 pages, of which 0 pages were retrieved with another vector.\n", "(50.00%).\n" ] }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
sentence_annotationid_annotationpage_annotationfile_annotation
200('r. Wir giengen in geschloffenen Reihen und m...a406cecb-48dc-4b27-9e2e-e831301a59bc149.0Z166069305_merged.txt
201('s. Da die Wagen nicht gewohnlich sind, so be...53e38871-8353-4ea1-acc5-e60964d4b02d151.0Z166069305_merged.txt
202('t. Man tritt mit den Fusen allerhand Muschel...32cc068f-6a60-4040-aa4a-0575f48b5c52155.0Z166069305_merged.txt
203('r. Sie kennen kein ander Spaziergehen, --- 0...638cd5aa-72b7-46a0-bbf1-82ec0f884747186.0Z166069305_merged.txt
204('s. In dem Departement, wo ich wohne und in d...8438c431-385f-40f0-b167-51527c2c8e7e221.0Z166069305_merged.txt
205('t. Ich wuste, das er besucht zu werden uerdi...41bd83f1-06ec-4186-af2e-63dea256dfc8267.0Z166069305_merged.txt
206('n. Sie hatten in ihrem Hause ein Paar rothe ...e5ebef45-e36c-4742-9976-702dfad43fc5270.0Z166069305_merged.txt
207('e. Nach diesem Kaufe nahmen wir uon dem gute...a02a872b-8602-4ff4-93bd-f539ff93ada6276.0Z166069305_merged.txt
208('n. Der Viceconsul, rein Drogman und ein fran...a5b12bfb-3a78-46ac-8d62-73d5bf4425c9277.0Z166069305_merged.txt
209('n. Unsere Esel Fielen nieder, sanken in den ...ceed7937-1824-4222-ba6c-686b2b4d7738278.0Z166069305_merged.txt
210('tu Wir hatten Maulefel, womit man gewohnlich...94d76c3d-1115-4a4e-91b5-501f6e88de65386.0Z166069305_merged.txt
211('. Ich trug kein Bedenken, mich ihnen anzuuer...acc36476-9315-4751-9646-bba3a2231e07386.0Z166069305_merged.txt
212('e. Damit man nun nicht an meiner Furchtlosig...b9449bfd-c3bb-4b97-9c45-305b10b7aeca389.0Z166069305_merged.txt
213('). Bei meiner Rukkunft nach Abukir fand ich ...995c9029-3950-44cb-a7c7-e6f91a96ecf7391.0Z166069305_merged.txt
214('-- Plinius berichtet nach dem Xenophon, die ...fe020016-1d51-4a1f-bbc1-d4a65973bf67402.0Z166069305_merged.txt
215('g. Ehe wir noch an dieseru Ore anlangten, sa...4795c52d-1f6e-4de6-b87f-a5ae17735dad419.0Z166069305_merged.txt
216('ort. Wir kamen auf einen\\' bedekten Sand, de...92f4ee0e-d411-4c55-bda6-02b5f8aefce6433.0Z166069305_merged.txt
217('n. Ob fie gleich in starkem Galop auf uns zu...0c529bb3-6ce7-4a2f-8741-1aaa4f2205d4434.0Z166069305_merged.txt
218('li Wenn man den Arabern glaubt, die der Mein...9a71ef03-d718-402b-b7f9-14844f486a22452.0Z166069305_merged.txt
219('t, Ich hatte einen uon den Landleuten, die i...e6e36f7b-3376-4624-a48b-ff2d88ba6167460.0Z166069305_merged.txt
220('91 Ich uerlies endlich diesen. heltischer Au...39769078-50ac-4faf-b32a-e74fd7cc46b0462.0Z166069305_merged.txt
\n", "
" ], "text/plain": [ " sentence_annotation \\\n", "200 ('r. Wir giengen in geschloffenen Reihen und m... \n", "201 ('s. Da die Wagen nicht gewohnlich sind, so be... \n", "202 ('t. Man tritt mit den Fusen allerhand Muschel... \n", "203 ('r. Sie kennen kein ander Spaziergehen, --- 0... \n", "204 ('s. In dem Departement, wo ich wohne und in d... \n", "205 ('t. Ich wuste, das er besucht zu werden uerdi... \n", "206 ('n. Sie hatten in ihrem Hause ein Paar rothe ... \n", "207 ('e. Nach diesem Kaufe nahmen wir uon dem gute... \n", "208 ('n. Der Viceconsul, rein Drogman und ein fran... \n", "209 ('n. Unsere Esel Fielen nieder, sanken in den ... \n", "210 ('tu Wir hatten Maulefel, womit man gewohnlich... \n", "211 ('. Ich trug kein Bedenken, mich ihnen anzuuer... \n", "212 ('e. Damit man nun nicht an meiner Furchtlosig... \n", "213 ('). Bei meiner Rukkunft nach Abukir fand ich ... \n", "214 ('-- Plinius berichtet nach dem Xenophon, die ... \n", "215 ('g. Ehe wir noch an dieseru Ore anlangten, sa... \n", "216 ('ort. Wir kamen auf einen\\' bedekten Sand, de... \n", "217 ('n. Ob fie gleich in starkem Galop auf uns zu... \n", "218 ('li Wenn man den Arabern glaubt, die der Mein... \n", "219 ('t, Ich hatte einen uon den Landleuten, die i... \n", "220 ('91 Ich uerlies endlich diesen. heltischer Au... \n", "\n", " id_annotation page_annotation \\\n", "200 a406cecb-48dc-4b27-9e2e-e831301a59bc 149.0 \n", "201 53e38871-8353-4ea1-acc5-e60964d4b02d 151.0 \n", "202 32cc068f-6a60-4040-aa4a-0575f48b5c52 155.0 \n", "203 638cd5aa-72b7-46a0-bbf1-82ec0f884747 186.0 \n", "204 8438c431-385f-40f0-b167-51527c2c8e7e 221.0 \n", "205 41bd83f1-06ec-4186-af2e-63dea256dfc8 267.0 \n", "206 e5ebef45-e36c-4742-9976-702dfad43fc5 270.0 \n", "207 a02a872b-8602-4ff4-93bd-f539ff93ada6 276.0 \n", "208 a5b12bfb-3a78-46ac-8d62-73d5bf4425c9 277.0 \n", "209 ceed7937-1824-4222-ba6c-686b2b4d7738 278.0 \n", "210 94d76c3d-1115-4a4e-91b5-501f6e88de65 386.0 \n", "211 acc36476-9315-4751-9646-bba3a2231e07 386.0 \n", "212 b9449bfd-c3bb-4b97-9c45-305b10b7aeca 389.0 \n", "213 995c9029-3950-44cb-a7c7-e6f91a96ecf7 391.0 \n", "214 fe020016-1d51-4a1f-bbc1-d4a65973bf67 402.0 \n", "215 4795c52d-1f6e-4de6-b87f-a5ae17735dad 419.0 \n", "216 92f4ee0e-d411-4c55-bda6-02b5f8aefce6 433.0 \n", "217 0c529bb3-6ce7-4a2f-8741-1aaa4f2205d4 434.0 \n", "218 9a71ef03-d718-402b-b7f9-14844f486a22 452.0 \n", "219 e6e36f7b-3376-4624-a48b-ff2d88ba6167 460.0 \n", "220 39769078-50ac-4faf-b32a-e74fd7cc46b0 462.0 \n", "\n", " file_annotation \n", "200 Z166069305_merged.txt \n", "201 Z166069305_merged.txt \n", "202 Z166069305_merged.txt \n", "203 Z166069305_merged.txt \n", "204 Z166069305_merged.txt \n", "205 Z166069305_merged.txt \n", "206 Z166069305_merged.txt \n", "207 Z166069305_merged.txt \n", "208 Z166069305_merged.txt \n", "209 Z166069305_merged.txt \n", "210 Z166069305_merged.txt \n", "211 Z166069305_merged.txt \n", "212 Z166069305_merged.txt \n", "213 Z166069305_merged.txt \n", "214 Z166069305_merged.txt \n", "215 Z166069305_merged.txt \n", "216 Z166069305_merged.txt \n", "217 Z166069305_merged.txt \n", "218 Z166069305_merged.txt \n", "219 Z166069305_merged.txt \n", "220 Z166069305_merged.txt " ] }, "execution_count": 10, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Filter rows where 'found_in_text_vector' is False --> False Negatives\n", "missed = matches_df[matches_df['found_in_text_vector'] == False]\n", "missed_fil = missed['id_annotation'].drop_duplicates(keep='first')\n", "missed_pages = missed['page_annotation'].drop_duplicates(keep='first')\n", "\n", "# Filter out pages that are in unique pages found\n", "missed_pages_unique = missed_pages[~missed_pages.isin(unique_foundPages_list)]\n", "retrieved_pages_otherVec = missed_pages[missed_pages.isin(unique_foundPages_list)]\n", "\n", "print(f\"False Negatives: Missed {len(missed_fil)} mentions on {len(missed_pages_unique)} pages, of which {len(retrieved_pages_otherVec)} pages were retrieved with another vector.\")\n", "print(f\"({(len(missed_pages_unique))/len(unique_pages_list)*100:.2f}%).\")\n", "\n", "missed[['sentence_annotation', 'id_annotation', 'page_annotation', 'file_annotation']]" ] }, { "cell_type": "code", "execution_count": 11, "id": "e7753ea4", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Retrieved 173 vectors on 173 pages of which 11 pages do contain annotations (correct pages, but other vector retrieved).\n", "and of which 162 pages do not contain annotations (= False Positives)\n", "(405.00)\n", "and of which 0 pages were retrieved with other vectors (= True Negatives).\n" ] }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
text_vector_xrerankid_text_vectorpagepage_annotationfile_annotation
16Unter einer Regierung, die nur die Absicht bat...17.09ee8949a-e0cc-4334-bc36-b71becd2d5c7400.0NaNNaN
23feiner Neise 1. th.).24.039e2e422-b7c3-40c1-90ee-12718c603261393.0NaNNaN
27**) Ursprung der Gefese 2. Bd.28.04ff5ac89-e2f3-468f-bad6-712ecce32e54170.0NaNNaN
28Von frih morgens an, bis auf den Abend, hat ma...29.0d4357d62-9022-4b84-803b-e634b791a923182.0NaNNaN
29Das Wasser, das * Strix passerina, Lin, Chevec...30.07aaae2e6-1e90-46a2-9f05-fe28c7b2842b236.0NaNNaN
.....................
195fchichte uberhaupt, und jene des. Mungo insbes...196.0f8995776-d337-413e-bc6b-d54a1fb5ff86223.0NaNNaN
196Ob sie gleich gesehen hatten, was seit Morgens...197.03315a574-24e7-4a73-b660-bd151a27956b437.0NaNNaN
197Die fehr oftern Uncinigkeiten zwischen den Beh...198.0bd45fa07-bd6a-46f3-9ba9-516d48d55fb5161.0NaNNaN
198Es ist bemerkenswerth, das man diesen unruhige...199.0736f2f9e-85b2-470d-a94e-f34f64db418989.0NaNNaN
199M. 24 und 26. Wit find Flein auf Erden und Flu...200.0ebb43751-cbfe-40d5-9056-95e68b9102a6126.0NaNNaN
\n", "

173 rows × 6 columns

\n", "
" ], "text/plain": [ " text_vector_x rerank \\\n", "16 Unter einer Regierung, die nur die Absicht bat... 17.0 \n", "23 feiner Neise 1. th.). 24.0 \n", "27 **) Ursprung der Gefese 2. Bd. 28.0 \n", "28 Von frih morgens an, bis auf den Abend, hat ma... 29.0 \n", "29 Das Wasser, das * Strix passerina, Lin, Chevec... 30.0 \n", ".. ... ... \n", "195 fchichte uberhaupt, und jene des. Mungo insbes... 196.0 \n", "196 Ob sie gleich gesehen hatten, was seit Morgens... 197.0 \n", "197 Die fehr oftern Uncinigkeiten zwischen den Beh... 198.0 \n", "198 Es ist bemerkenswerth, das man diesen unruhige... 199.0 \n", "199 M. 24 und 26. Wit find Flein auf Erden und Flu... 200.0 \n", "\n", " id_text_vector page page_annotation \\\n", "16 9ee8949a-e0cc-4334-bc36-b71becd2d5c7 400.0 NaN \n", "23 39e2e422-b7c3-40c1-90ee-12718c603261 393.0 NaN \n", "27 4ff5ac89-e2f3-468f-bad6-712ecce32e54 170.0 NaN \n", "28 d4357d62-9022-4b84-803b-e634b791a923 182.0 NaN \n", "29 7aaae2e6-1e90-46a2-9f05-fe28c7b2842b 236.0 NaN \n", ".. ... ... ... \n", "195 f8995776-d337-413e-bc6b-d54a1fb5ff86 223.0 NaN \n", "196 3315a574-24e7-4a73-b660-bd151a27956b 437.0 NaN \n", "197 bd45fa07-bd6a-46f3-9ba9-516d48d55fb5 161.0 NaN \n", "198 736f2f9e-85b2-470d-a94e-f34f64db4189 89.0 NaN \n", "199 ebb43751-cbfe-40d5-9056-95e68b9102a6 126.0 NaN \n", "\n", " file_annotation \n", "16 NaN \n", "23 NaN \n", "27 NaN \n", "28 NaN \n", "29 NaN \n", ".. ... \n", "195 NaN \n", "196 NaN \n", "197 NaN \n", "198 NaN \n", "199 NaN \n", "\n", "[173 rows x 6 columns]" ] }, "execution_count": 11, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Filter rows where 'found_in_text_vector' is NaN --> False Positives\n", "FP = matches_df[matches_df['found_in_text_vector'].isna()]\n", "fP_pages = FP['page'].drop_duplicates(keep='first')\n", "\n", "# Extract the \"page\" value from each dictionary of unique pages found\n", "unique_FPPages_list = fP_pages.to_list()\n", "\n", "# Filter out pages that are in unique pages found\n", "ret_pages_Annot = fP_pages[fP_pages.isin(unique_pages_list)]\n", "ret_pages_noAnnot = fP_pages[~fP_pages.isin(unique_pages_list)]\n", "retrieved_otherVec = fP_pages[fP_pages.isin(unique_foundPages_list)]\n", "\n", "print(f\"Retrieved {len(FP)} vectors on {len(unique_FPPages_list)} pages of which {len(ret_pages_Annot)} pages do contain annotations (correct pages, but other vector retrieved).\")\n", "print(f\"and of which {len(ret_pages_noAnnot)} pages do not contain annotations (= False Positives)\")\n", "print(f\"({len(ret_pages_noAnnot)/len(unique_pages_list)*100:.2f})\")\n", "print(f\"and of which {len(retrieved_otherVec)} pages were retrieved with other vectors (= True Negatives).\")\n", "FP[['text_vector_x', 'rerank', 'id_text_vector', 'page', 'page_annotation', 'file_annotation']]" ] }, { "cell_type": "code", "execution_count": null, "id": "2bf572ee", "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3 (ipykernel)", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.9.12" }, "toc": { "base_numbering": 1, "nav_menu": {}, "number_sections": true, "sideBar": true, "skip_h1_title": false, "title_cell": "Table of Contents", "title_sidebar": "Contents", "toc_cell": false, "toc_position": {}, "toc_section_display": true, "toc_window_display": false } }, "nbformat": 4, "nbformat_minor": 5 }