{"id":165995,"date":"2014-02-24T00:00:00","date_gmt":"2014-02-24T00:00:00","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/msr-research-item\/entity-linking-at-the-tail-sparse-signals-unknown-entities-and-phrase-models\/"},"modified":"2018-10-16T20:04:44","modified_gmt":"2018-10-17T03:04:44","slug":"entity-linking-at-the-tail-sparse-signals-unknown-entities-and-phrase-models","status":"publish","type":"msr-research-item","link":"https:\/\/www.microsoft.com\/en-us\/research\/publication\/entity-linking-at-the-tail-sparse-signals-unknown-entities-and-phrase-models\/","title":{"rendered":"Entity Linking at the Tail: Sparse Signals, Unknown Entities and Phrase Models"},"content":{"rendered":"<div class=\"asset-content\">\n<p>Web search is seeing a paradigm shift from keyword based search to an entity-centric organization of web data. To support web search with this deeper level of understanding, a web-scale entity linking system must include feature extraction that is robust to the diversity of web documents and their varied writing styles and content structure; maintain high-precision linking for &#8220;tail&#8221; (unpopular) entities that is robust to the existence of confounding entities outside of the knowledge base and entity profiles with minimal information; and represent large-scale knowledge bases with a scalable and powerful feature representation. We have built and deployed a web-scale unsupervised entity linking system for a commercial search engine that addresses these requirements by combining new developments in sparse signal recovery to identify the most discriminative features from noisy, free-text web documents; explicit modeling of out-of-knowledge-base entities to improve precision at the tail; and the development of a new phrase-unigram language model to efficiently capture high-order dependencies in lexical features. Using a knowledge base of 100M unique people from a popular social networking site, we present experimental results in the challenging domain of people-linking at the tail, where most entities have limited web presence. Our experimental results show that this system substantially improves on the precision-recall tradeoff over baseline methods, achieving precision over 95% with recall over 60%.<\/p>\n<\/div>\n<p><!-- .asset-content --><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Web search is seeing a paradigm shift from keyword based search to an entity-centric organization of web data. To support web search with this deeper level of understanding, a web-scale entity linking system must include feature extraction that is robust to the diversity of web documents and their varied writing styles and content structure; maintain [&hellip;]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","msr-author-ordering":[{"type":"user_nicename","value":"yuzjin"},{"type":"user_nicename","value":"emrek"},{"type":"user_nicename","value":"kuansanw"},{"type":"user_nicename","value":"riloynd"}],"msr_publishername":"ACM","msr_publisher_other":"","msr_booktitle":"","msr_chapter":"","msr_edition":"WSDM '14 Proceedings of the 7th ACM international conference on Web search and data mining","msr_editors":"","msr_how_published":"","msr_isbn":"","msr_issue":"","msr_journal":"","msr_number":"","msr_organization":"","msr_pages_string":"Pages 453-462","msr_page_range_start":"453","msr_page_range_end":"462","msr_series":"","msr_volume":"","msr_copyright":"","msr_conference_name":"WSDM '14 Proceedings of the 7th ACM international conference on Web search and data mining","msr_doi":"10.1145\/2556195.2556230","msr_arxiv_id":"","msr_s2_paper_id":"","msr_mag_id":"","msr_pubmed_id":"","msr_other_authors":"","msr_other_contributors":"","msr_speaker":"","msr_award":"","msr_affiliation":"","msr_institution":"","msr_host":"","msr_version":"","msr_duration":"","msr_original_fields_of_study":"","msr_release_tracker_id":"","msr_s2_match_type":"","msr_citation_count_updated":"","msr_published_date":"2014-02-24","msr_highlight_text":"","msr_notes":"","msr_longbiography":"","msr_publicationurl":"","msr_external_url":"","msr_secondary_video_url":"","msr_conference_url":"","msr_journal_url":"","msr_s2_pdf_url":"","msr_year":2014,"msr_citation_count":0,"msr_influential_citations":0,"msr_reference_count":0,"msr_s2_match_confidence":0,"msr_microsoftintellectualproperty":true,"msr_s2_open_access":false,"msr_s2_author_ids":[],"msr_pub_ids":[],"msr_hide_image_in_river":0,"footnotes":""},"msr-research-highlight":[],"research-area":[13555],"msr-publication-type":[193716],"msr-publisher":[],"msr-focus-area":[],"msr-locale":[268875],"msr-post-option":[],"msr-field-of-study":[],"msr-conference":[],"msr-journal":[],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-165995","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-search-information-retrieval","msr-locale-en_us"],"msr_publishername":"ACM","msr_edition":"WSDM '14 Proceedings of the 7th ACM international conference on Web search and data mining","msr_affiliation":"","msr_published_date":"2014-02-24","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"Pages 453-462","msr_chapter":"","msr_isbn":"","msr_journal":"","msr_volume":"","msr_number":"","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"","msr_how_published":"","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":1,"msr_main_download":"205045","msr_publicationurl":"","msr_doi":"10.1145\/2556195.2556230","msr_publication_uploader":[{"type":"file","title":"wsdm180-jin.pdf","viewUrl":"https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2016\/02\/wsdm180-jin.pdf","id":205045,"label_id":0},{"type":"doi","title":"10.1145\/2556195.2556230","viewUrl":false,"id":false,"label_id":0}],"msr_related_uploader":"","msr_citation_count":0,"msr_citation_count_updated":"","msr_s2_paper_id":"","msr_influential_citations":0,"msr_reference_count":0,"msr_arxiv_id":"","msr_s2_author_ids":[],"msr_s2_open_access":false,"msr_s2_pdf_url":null,"msr_attachments":[{"id":205045,"url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2016\/02\/wsdm180-jin.pdf"}],"msr-author-ordering":[{"type":"user_nicename","value":"yuzjin","user_id":35089,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=yuzjin"},{"type":"user_nicename","value":"emrek","user_id":31739,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=emrek"},{"type":"user_nicename","value":"kuansanw","user_id":32592,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=kuansanw"},{"type":"user_nicename","value":"riloynd","user_id":33406,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=riloynd"}],"msr_impact_theme":[],"msr_research_lab":[],"msr_event":[],"msr_group":[144672],"msr_project":[],"publication":[],"video":[],"msr-tool":[],"msr_publication_type":"inproceedings","related_content":[],"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/165995","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":1,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/165995\/revisions"}],"predecessor-version":[{"id":521840,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/165995\/revisions\/521840"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=165995"}],"wp:term":[{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=165995"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=165995"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=165995"},{"taxonomy":"msr-publisher","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-publisher?post=165995"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=165995"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=165995"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=165995"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=165995"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=165995"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=165995"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=165995"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=165995"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}