{"id":{"repo_id":"arizona-thes","oai_identifier":"oai:repository.arizona.edu:10150/666175"},"canonical_url":"https://search.dev.ndltd.org/etd/arizona-thes/oai:repository.arizona.edu:10150/666175","repository":{"repo_id":"arizona-thes","name":"University of Arizona","base_url":"https://repository.arizona.edu/oai/request"},"display":{"title":"Enhanced Recommender Systems by Biclustering","abstract":"Recommender systems play a very important role to explore and suggest personalized recommendations to users from a huge number of choices. In this thesis, we propose a new class of enhanced recommender systems, called BiRDS (meaning Biclustering Recommendation Systems) and develop three different modeling approaches under the new framework: iBiRDS (imputation BiRDS), rBiRDS (regularization BiRDS), lBiRDS (logistic BiRDS). All the three methods utilize biclustering to incorporate correlational traits between users and items. In this research work, we propose to identify the biclusters of associated users and items through a weighted bipartite network of users' feedback and a fast community detection algorithm. Concurrent information of users and items is either imputed or integrated into the singular value decomposition (SVD) framework without requiring any domain knowledge. Computationally, the BiRDS estimation avoids large matrices operation and memory storage, making it advantageous to attain scalability for massive datasets. Moreover, the BiRDS can effectively combat the ``cold-start\" issue by utilizing bicluster effects of associated users and items while most collaborative filtering methods depend on subject-specific parameters only. For the proposed BiRDS methods, we develop their estimation framework and computational algorithms, study their computational complexity and convergence properties, and establish their theoretical guarantee in terms of the asymptotic properties. In addition, we perform extensive numerical experiments to evaluate the performance of the new methods and compare them with existing techniques in the literature. Our simulation studies and several real data analysis demonstrate that the proposed methods can enhance prediction accuracy of recommender systems with significantly less computing time compared to existing competitive approaches.","abstract_html":"Recommender systems play a very important role to explore and suggest personalized recommendations to users from a huge number of choices. In this thesis, we propose a new class of enhanced recommender systems, called BiRDS (meaning Biclustering Recommendation Systems) and develop three different modeling approaches under the new framework: iBiRDS (imputation BiRDS), rBiRDS (regularization BiRDS), lBiRDS (logistic BiRDS). All the three methods utilize biclustering to incorporate correlational traits between users and items. In this research work, we propose to identify the biclusters of associated users and items through a weighted bipartite network of users' feedback and a fast community detection algorithm. Concurrent information of users and items is either imputed or integrated into the singular value decomposition (SVD) framework without requiring any domain knowledge. Computationally, the BiRDS estimation avoids large matrices operation and memory storage, making it advantageous to attain scalability for massive datasets. Moreover, the BiRDS can effectively combat the ``cold-start" issue by utilizing bicluster effects of associated users and items while most collaborative filtering methods depend on subject-specific parameters only. For the proposed BiRDS methods, we develop their estimation framework and computational algorithms, study their computational complexity and convergence properties, and establish their theoretical guarantee in terms of the asymptotic properties. In addition, we perform extensive numerical experiments to evaluate the performance of the new methods and compare them with existing techniques in the literature. Our simulation studies and several real data analysis demonstrate that the proposed methods can enhance prediction accuracy of recommender systems with significantly less computing time compared to existing competitive approaches.","abstract_has_math":false,"creators":["Chung, Kyung Mi"],"institution":"The University of Arizona.","degree_name":"Ph.D.","degree_level":"doctoral","degree_discipline":"Graduate College","degree_department":null,"school":null,"contributors":[],"advisors":["Zhang, Hao"],"committee_chairs":[],"committee_members":["Hao, Ning","Fan, Neng","Tang, Xueying"],"year":2022,"date_issued":"2022","date_published":"2022","updated_at":"2026-07-24T00:55:59Z","subjects":[],"languages":["en"],"rights":["Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction, presentation (such as public display or performance) of protected items is prohibited except with permission of the author."],"rights_urls":["http://rightsstatements.org/vocab/InC/1.0/"],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/10150/666175","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Zhang, Hao"]},{"key":"dc:contributor.committeemember","label":"Committee Member","values":["Hao, Ning","Fan, Neng","Tang, Xueying"]},{"key":"dc:creator","label":"Author","values":["Chung, Kyung Mi"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2022-09-22T01:33:11Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2022-09-22T01:33:11Z"]},{"key":"dc:date.issued","label":"Date","values":["2022"]},{"key":"dc:publisher","label":"Institution","values":["The University of Arizona."]},{"key":"dc:type","label":"Dc Type","values":["text","Electronic Dissertation"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Graduate College","Statistics"]},{"key":"thesis:degree_level","label":"Degree Level","values":["doctoral"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Ph.D."]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["University of Arizona"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction, presentation (such as public display or performance) of protected items is prohibited except with permission of the author."]},{"key":"dc:rights.uri","label":"Rights URI","values":["http://rightsstatements.org/vocab/InC/1.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["http://hdl.handle.net/10150/666175"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Recommender systems play a very important role to explore and suggest personalized recommendations to users from a huge number of choices. In this thesis, we propose a new class of enhanced recommender systems, called BiRDS (meaning Biclustering Recommendation Systems) and develop three different modeling approaches under the new framework: iBiRDS (imputation BiRDS), rBiRDS (regularization BiRDS), lBiRDS (logistic BiRDS). All the three methods utilize biclustering to incorporate correlational traits between users and items. In this research work, we propose to identify the biclusters of associated users and items through a weighted bipartite network of users' feedback and a fast community detection algorithm. Concurrent information of users and items is either imputed or integrated into the singular value decomposition (SVD) framework without requiring any domain knowledge. Computationally, the BiRDS estimation avoids large matrices operation and memory storage, making it advantageous to attain scalability for massive datasets. Moreover, the BiRDS can effectively combat the ``cold-start\" issue by utilizing bicluster effects of associated users and items while most collaborative filtering methods depend on subject-specific parameters only. For the proposed BiRDS methods, we develop their estimation framework and computational algorithms, study their computational complexity and convergence properties, and establish their theoretical guarantee in terms of the asymptotic properties. In addition, we perform extensive numerical experiments to evaluate the performance of the new methods and compare them with existing techniques in the literature. Our simulation studies and several real data analysis demonstrate that the proposed methods can enhance prediction accuracy of recommender systems with significantly less computing time compared to existing competitive approaches."]},{"key":"dc:title","label":"Title","values":["Enhanced Recommender Systems by Biclustering"]}]}],"canonical_facts":{"dc:contributor.advisor":["Zhang, Hao"],"dc:contributor.committeemember":["Hao, Ning","Fan, Neng","Tang, Xueying"],"dc:creator":["Chung, Kyung Mi"],"dc:date.accessioned":["2022-09-22T01:33:11Z"],"dc:date.available":["2022-09-22T01:33:11Z"],"dc:date.issued":["2022"],"dc:description.abstract":["Recommender systems play a very important role to explore and suggest personalized recommendations to users from a huge number of choices. In this thesis, we propose a new class of enhanced recommender systems, called BiRDS (meaning Biclustering Recommendation Systems) and develop three different modeling approaches under the new framework: iBiRDS (imputation BiRDS), rBiRDS (regularization BiRDS), lBiRDS (logistic BiRDS). All the three methods utilize biclustering to incorporate correlational traits between users and items. In this research work, we propose to identify the biclusters of associated users and items through a weighted bipartite network of users' feedback and a fast community detection algorithm. Concurrent information of users and items is either imputed or integrated into the singular value decomposition (SVD) framework without requiring any domain knowledge. Computationally, the BiRDS estimation avoids large matrices operation and memory storage, making it advantageous to attain scalability for massive datasets. Moreover, the BiRDS can effectively combat the ``cold-start\" issue by utilizing bicluster effects of associated users and items while most collaborative filtering methods depend on subject-specific parameters only. For the proposed BiRDS methods, we develop their estimation framework and computational algorithms, study their computational complexity and convergence properties, and establish their theoretical guarantee in terms of the asymptotic properties. In addition, we perform extensive numerical experiments to evaluate the performance of the new methods and compare them with existing techniques in the literature. Our simulation studies and several real data analysis demonstrate that the proposed methods can enhance prediction accuracy of recommender systems with significantly less computing time compared to existing competitive approaches."],"dc:identifier.uri":["http://hdl.handle.net/10150/666175"],"dc:language.iso":["en"],"dc:publisher":["The University of Arizona."],"dc:rights":["Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction, presentation (such as public display or performance) of protected items is prohibited except with permission of the author."],"dc:rights.uri":["http://rightsstatements.org/vocab/InC/1.0/"],"dc:title":["Enhanced Recommender Systems by Biclustering"],"dc:type":["text","Electronic Dissertation"],"thesis:degree_discipline":["Graduate College","Statistics"],"thesis:degree_level":["doctoral"],"thesis:degree_name":["Ph.D."],"thesis:institution_name":["University of Arizona"]},"updated_at":"2026-07-24T00:55:59Z"}