{"id":{"repo_id":"uiuc","oai_identifier":"oai:www.ideals.illinois.edu:2142/92709"},"canonical_url":"https://search.dev.ndltd.org/etd/uiuc/oai:www.ideals.illinois.edu:2142/92709","repository":{"repo_id":"uiuc","name":"University of Illinois - Urbana-Champaign","base_url":"https://www.ideals.illinois.edu/oai-pmh"},"display":{"title":"Axiomatic analysis of smoothing methods in language models for pseudo-relevance feedback","abstract":"Pseudo-Relevance Feedback (PRF) is an important general technique for improving retrieval effectiveness without requiring any user effort. Several state-of-the-art PRF models are based on the language modeling approach where a query language model is learned based on feedback documents. In all these models, feedback documents are represented with unigram language models smoothed with a collection language model. While collection language model-based smoothing has proven both effective and necessary in using language models for retrieval, we use axiomatic analysis to show that this smoothing scheme inherently causes the feedback model to favor frequent terms and thus violates the IDF constraint needed to ensure selection of discriminative feedback terms. To address this problem, we propose replacing collection language model-based smoothing in the feedback stage with additive smoothing, which is analytically shown to select more discriminative terms. Empirical evaluation further confirms that additive smoothing indeed significantly outperforms collection-based smoothing methods in multiple language model-based PRF models.","abstract_html":"Pseudo-Relevance Feedback (PRF) is an important general technique for improving retrieval effectiveness without requiring any user effort. Several state-of-the-art PRF models are based on the language modeling approach where a query language model is learned based on feedback documents. In all these models, feedback documents are represented with unigram language models smoothed with a collection language model. While collection language model-based smoothing has proven both effective and necessary in using language models for retrieval, we use axiomatic analysis to show that this smoothing scheme inherently causes the feedback model to favor frequent terms and thus violates the IDF constraint needed to ensure selection of discriminative feedback terms. To address this problem, we propose replacing collection language model-based smoothing in the feedback stage with additive smoothing, which is analytically shown to select more discriminative terms. Empirical evaluation further confirms that additive smoothing indeed significantly outperforms collection-based smoothing methods in multiple language model-based PRF models.","abstract_has_math":false,"creators":["Hazimeh, Hussein"],"institution":"University of Illinois at Urbana-Champaign","degree_name":"M.S.","degree_level":"Thesis","degree_discipline":"Computer Science","degree_department":null,"school":null,"contributors":["Zhai, ChengXiang"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2016,"date_issued":"2016-11-10T17:49:42Z","date_published":"2016-11-10T17:49:42Z","updated_at":"2026-07-22T22:26:35Z","subjects":["Search Engines","Text Retrieval","Relevance Feedback","Pseudo-Relevance Feedback","Implicit Feedback","Blind Feedback"],"languages":["en"],"rights":["Copyright 2016 Hussein Hazimeh"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/2142/92709","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Zhai, ChengXiang"]},{"key":"dc:creator","label":"Author","values":["Hazimeh, Hussein"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2016-11-10T17:49:42Z","2016-06-02","2016-08"]},{"key":"dc:type","label":"Dc Type","values":["text"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Computer Science"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Thesis"]},{"key":"thesis:degree_name","label":"Degree Name","values":["M.S."]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["University of Illinois at Urbana-Champaign"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Search Engines","Text Retrieval","Relevance Feedback","Pseudo-Relevance Feedback","Implicit Feedback","Blind Feedback"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["Copyright 2016 Hussein Hazimeh"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["http://hdl.handle.net/2142/92709"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Pseudo-Relevance Feedback (PRF) is an important general technique for improving retrieval effectiveness without requiring any user effort. Several state-of-the-art PRF models are based on the language modeling approach where a query language model is learned based on feedback documents. In all these models, feedback documents are represented with unigram language models smoothed with a collection language model. While collection language model-based smoothing has proven both effective and necessary in using language models for retrieval, we use axiomatic analysis to show that this smoothing scheme inherently causes the feedback model to favor frequent terms and thus violates the IDF constraint needed to ensure selection of discriminative feedback terms. To address this problem, we propose replacing collection language model-based smoothing in the feedback stage with additive smoothing, which is analytically shown to select more discriminative terms. Empirical evaluation further confirms that additive smoothing indeed significantly outperforms collection-based smoothing methods in multiple language model-based PRF models.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2016-11-09 without embargo terms","The student, Hussein Hazimeh, accepted the attached license on 2016-06-02 at 00:51.","The student, Hussein Hazimeh, submitted this Thesis for approval on 2016-06-02 at 00:58.","This Thesis was approved for publication on 2016-06-02 at 11:26.","DSpace SAF Submission Ingestion Package generated from Vireo submission #9637 on 2016-11-09 at 10:20:59","Made available in DSpace on 2016-11-10T17:49:42Z (GMT). No. of bitstreams: 2 HAZIMEH-THESIS-2016.pdf: 590964 bytes, checksum: 3feaaeb46c714aa129bfbc3d77326951 (MD5) LICENSE.txt: 4212 bytes, checksum: 2e2cf52d36fddc987988385a0421dbf3 (MD5) Previous issue date: 2016-06-02"]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Axiomatic analysis of smoothing methods in language models for pseudo-relevance feedback"]}]}],"canonical_facts":{"dc:contributor":["Zhai, ChengXiang"],"dc:creator":["Hazimeh, Hussein"],"dc:date":["2016-11-10T17:49:42Z","2016-06-02","2016-08"],"dc:description":["Pseudo-Relevance Feedback (PRF) is an important general technique for improving retrieval effectiveness without requiring any user effort. Several state-of-the-art PRF models are based on the language modeling approach where a query language model is learned based on feedback documents. In all these models, feedback documents are represented with unigram language models smoothed with a collection language model. While collection language model-based smoothing has proven both effective and necessary in using language models for retrieval, we use axiomatic analysis to show that this smoothing scheme inherently causes the feedback model to favor frequent terms and thus violates the IDF constraint needed to ensure selection of discriminative feedback terms. To address this problem, we propose replacing collection language model-based smoothing in the feedback stage with additive smoothing, which is analytically shown to select more discriminative terms. Empirical evaluation further confirms that additive smoothing indeed significantly outperforms collection-based smoothing methods in multiple language model-based PRF models.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2016-11-09 without embargo terms","The student, Hussein Hazimeh, accepted the attached license on 2016-06-02 at 00:51.","The student, Hussein Hazimeh, submitted this Thesis for approval on 2016-06-02 at 00:58.","This Thesis was approved for publication on 2016-06-02 at 11:26.","DSpace SAF Submission Ingestion Package generated from Vireo submission #9637 on 2016-11-09 at 10:20:59","Made available in DSpace on 2016-11-10T17:49:42Z (GMT). No. of bitstreams: 2 HAZIMEH-THESIS-2016.pdf: 590964 bytes, checksum: 3feaaeb46c714aa129bfbc3d77326951 (MD5) LICENSE.txt: 4212 bytes, checksum: 2e2cf52d36fddc987988385a0421dbf3 (MD5) Previous issue date: 2016-06-02"],"dc:format":["application/pdf"],"dc:identifier":["http://hdl.handle.net/2142/92709"],"dc:language":["en"],"dc:rights":["Copyright 2016 Hussein Hazimeh"],"dc:subject":["Search Engines","Text Retrieval","Relevance Feedback","Pseudo-Relevance Feedback","Implicit Feedback","Blind Feedback"],"dc:title":["Axiomatic analysis of smoothing methods in language models for pseudo-relevance feedback"],"dc:type":["text"],"thesis:degree_discipline":["Computer Science"],"thesis:degree_level":["Thesis"],"thesis:degree_name":["M.S."],"thesis:institution_name":["University of Illinois at Urbana-Champaign"]},"updated_at":"2026-07-22T22:26:35Z"}