Instilling Type Knowledge in Language Models via Multi-Task QA

Understanding human language often necessitates understanding entities and their place in a taxonomy of knowledge -- their types. Previous methods to learn entity types rely on training classifiers on datasets with coarse, noisy, and incomplete labels. We introduce a method to instill fine-grained t...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2022-04
Main Authors: Li, Shuyang, Sridhar, Mukund, Prakash, Chandana Satya, Cao, Jin, Hamza, Wael, McAuley, Julian
Format: Article
Language:English
Subjects:
Datasets
Encyclopedias
Knowledge
Knowledge bases (artificial intelligence)
Knowledge representation
Taxonomy
Training
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
cited_by
cites
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Li, Shuyang
Sridhar, Mukund
Prakash, Chandana Satya
Cao, Jin
Hamza, Wael
McAuley, Julian
description Understanding human language often necessitates understanding entities and their place in a taxonomy of knowledge -- their types. Previous methods to learn entity types rely on training classifiers on datasets with coarse, noisy, and incomplete labels. We introduce a method to instill fine-grained type knowledge in language models with text-to-text pre-training on type-centric questions leveraging knowledge base documents and knowledge graphs. We create the WikiWiki dataset: entities and passages from 10M Wikipedia articles linked to the Wikidata knowledge graph with 41K types. Models trained on WikiWiki achieve state-of-the-art performance in zero-shot dialog state tracking benchmarks, accurately infer entity types in Wikipedia articles, and can discover new types deemed useful by human judges.
format article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2658422107</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2658422107</sourcerecordid><originalsourceid>FETCH-proquest_journals_26584221073</originalsourceid><addsrcrecordid>eNpjYuA0MjY21LUwMTLiYOAtLs4yMDAwMjM3MjU15mSw88wrLsnMycnMS1cIqSxIVfDOyy_PSU1JT1XIzFPwScxLL00Esn3zU1JzihXKMhMVfEtzSjJ1QxKLsxUCHXkYWNMSc4pTeaE0N4Oym2uIs4duQVF-YWlqcUl8Vn5pUR5QKt7IzBTkBEMDc2PiVAEAY6I3aA</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2658422107</pqid></control><display><type>article</type><title>Instilling Type Knowledge in Language Models via Multi-Task QA</title><source>Publicly Available Content Database</source><creator>Li, Shuyang ; Sridhar, Mukund ; Prakash, Chandana Satya ; Cao, Jin ; Hamza, Wael ; McAuley, Julian</creator><creatorcontrib>Li, Shuyang ; Sridhar, Mukund ; Prakash, Chandana Satya ; Cao, Jin ; Hamza, Wael ; McAuley, Julian</creatorcontrib><description>Understanding human language often necessitates understanding entities and their place in a taxonomy of knowledge -- their types. Previous methods to learn entity types rely on training classifiers on datasets with coarse, noisy, and incomplete labels. We introduce a method to instill fine-grained type knowledge in language models with text-to-text pre-training on type-centric questions leveraging knowledge base documents and knowledge graphs. We create the WikiWiki dataset: entities and passages from 10M Wikipedia articles linked to the Wikidata knowledge graph with 41K types. Models trained on WikiWiki achieve state-of-the-art performance in zero-shot dialog state tracking benchmarks, accurately infer entity types in Wikipedia articles, and can discover new types deemed useful by human judges.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Datasets ; Encyclopedias ; Knowledge ; Knowledge bases (artificial intelligence) ; Knowledge representation ; Taxonomy ; Training</subject><ispartof>arXiv.org, 2022-04</ispartof><rights>2022. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://www.proquest.com/docview/2658422107?pq-origsite=primo$$EHTML$$P50$$Gproquest$$Hfree_for_read</linktohtml><link.rule.ids>780,784,25752,37011,44589</link.rule.ids></links><search><creatorcontrib>Li, Shuyang</creatorcontrib><creatorcontrib>Sridhar, Mukund</creatorcontrib><creatorcontrib>Prakash, Chandana Satya</creatorcontrib><creatorcontrib>Cao, Jin</creatorcontrib><creatorcontrib>Hamza, Wael</creatorcontrib><creatorcontrib>McAuley, Julian</creatorcontrib><title>Instilling Type Knowledge in Language Models via Multi-Task QA</title><title>arXiv.org</title><description>Understanding human language often necessitates understanding entities and their place in a taxonomy of knowledge -- their types. Previous methods to learn entity types rely on training classifiers on datasets with coarse, noisy, and incomplete labels. We introduce a method to instill fine-grained type knowledge in language models with text-to-text pre-training on type-centric questions leveraging knowledge base documents and knowledge graphs. We create the WikiWiki dataset: entities and passages from 10M Wikipedia articles linked to the Wikidata knowledge graph with 41K types. Models trained on WikiWiki achieve state-of-the-art performance in zero-shot dialog state tracking benchmarks, accurately infer entity types in Wikipedia articles, and can discover new types deemed useful by human judges.</description><subject>Datasets</subject><subject>Encyclopedias</subject><subject>Knowledge</subject><subject>Knowledge bases (artificial intelligence)</subject><subject>Knowledge representation</subject><subject>Taxonomy</subject><subject>Training</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><sourceid>PIMPY</sourceid><recordid>eNpjYuA0MjY21LUwMTLiYOAtLs4yMDAwMjM3MjU15mSw88wrLsnMycnMS1cIqSxIVfDOyy_PSU1JT1XIzFPwScxLL00Esn3zU1JzihXKMhMVfEtzSjJ1QxKLsxUCHXkYWNMSc4pTeaE0N4Oym2uIs4duQVF-YWlqcUl8Vn5pUR5QKt7IzBTkBEMDc2PiVAEAY6I3aA</recordid><startdate>20220428</startdate><enddate>20220428</enddate><creator>Li, Shuyang</creator><creator>Sridhar, Mukund</creator><creator>Prakash, Chandana Satya</creator><creator>Cao, Jin</creator><creator>Hamza, Wael</creator><creator>McAuley, Julian</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope></search><sort><creationdate>20220428</creationdate><title>Instilling Type Knowledge in Language Models via Multi-Task QA</title><author>Li, Shuyang ; Sridhar, Mukund ; Prakash, Chandana Satya ; Cao, Jin ; Hamza, Wael ; McAuley, Julian</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_26584221073</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Datasets</topic><topic>Encyclopedias</topic><topic>Knowledge</topic><topic>Knowledge bases (artificial intelligence)</topic><topic>Knowledge representation</topic><topic>Taxonomy</topic><topic>Training</topic><toplevel>online_resources</toplevel><creatorcontrib>Li, Shuyang</creatorcontrib><creatorcontrib>Sridhar, Mukund</creatorcontrib><creatorcontrib>Prakash, Chandana Satya</creatorcontrib><creatorcontrib>Cao, Jin</creatorcontrib><creatorcontrib>Hamza, Wael</creatorcontrib><creatorcontrib>McAuley, Julian</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni)</collection><collection>ProQuest Central</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central Korea</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Li, Shuyang</au><au>Sridhar, Mukund</au><au>Prakash, Chandana Satya</au><au>Cao, Jin</au><au>Hamza, Wael</au><au>McAuley, Julian</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Instilling Type Knowledge in Language Models via Multi-Task QA</atitle><jtitle>arXiv.org</jtitle><date>2022-04-28</date><risdate>2022</risdate><eissn>2331-8422</eissn><abstract>Understanding human language often necessitates understanding entities and their place in a taxonomy of knowledge -- their types. Previous methods to learn entity types rely on training classifiers on datasets with coarse, noisy, and incomplete labels. We introduce a method to instill fine-grained type knowledge in language models with text-to-text pre-training on type-centric questions leveraging knowledge base documents and knowledge graphs. We create the WikiWiki dataset: entities and passages from 10M Wikipedia articles linked to the Wikidata knowledge graph with 41K types. Models trained on WikiWiki achieve state-of-the-art performance in zero-shot dialog state tracking benchmarks, accurately infer entity types in Wikipedia articles, and can discover new types deemed useful by human judges.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2022-04
issn 2331-8422
language eng
recordid cdi_proquest_journals_2658422107
source Publicly Available Content Database
subjects Datasets
Encyclopedias
Knowledge
Knowledge bases (artificial intelligence)
Knowledge representation
Taxonomy
Training
title Instilling Type Knowledge in Language Models via Multi-Task QA
url http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-08T18%3A24%3A23IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=Instilling%20Type%20Knowledge%20in%20Language%20Models%20via%20Multi-Task%20QA&rft.jtitle=arXiv.org&rft.au=Li,%20Shuyang&rft.date=2022-04-28&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2658422107%3C/proquest%3E%3Cgrp_id%3Ecdi_FETCH-proquest_journals_26584221073%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=2658422107&rft_id=info:pmid/&rfr_iscdi=true