Factored MDPs for detecting topics of user sessions

Research output: Contributions to collected editions/worksArticle in conference proceedingsResearchpeer-review

Authors

Recommender systems aim to capture interests of users to provide tailored recommendations. User interests are however often unique and depend on many unobservable factors including a user's mood and the local weather. We take a contextual session-based approach and propose a sequential framework using factored Markov decision processes (fMDPs) to detect the user's goal (the topic) of a session. We show that an independence assumption on the attributes of items leads to a set of independent models that can be optimised efficiently. Our approach results in interpretable topics that can be effectively turned into recommendations. Empirical results on a real world click log from a large e-commerce company exhibit highly accurate topic prediction rates of about 90%. Translating our approach into a topic-driven recommender system outperforms several baseline competitors.

Original languageEnglish
Title of host publicationRecSys 2014 - Proceedings of the 8th ACM Conference on Recommender Systems
Number of pages8
PublisherAssociation for Computing Machinery, Inc
Publication date06.10.2014
Pages33-40
ISBN (print)978-1-4503-2668-1
DOIs
Publication statusPublished - 06.10.2014
Externally publishedYes
Event8th ACM Conference on Recommender Systems - RecSys2014 - Crowne Plaza hotel in Foster City, Foster City, United States
Duration: 06.10.201410.10.2014
Conference number: 8
https://recsys.acm.org/recsys14/

DOI