PathMining: A path-based user profiling algorithm for heterogeneous graph-based recommender systems

Sangkeun Lee, Sanghyeb Lee, Byoung Hoon Park

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

5 Scopus citations

Abstract

In a heterogeneous graph-based recommender system, relationships among various entities are used to predict the rating or preference that a user would give to an item. By modeling user's selection behavior as a path in the heterogeneous graph, we can capture each user's unique selection behaviors as a set of paths with specific types of nodes and edges. Since these paths capture selection behaviors unique to each user, these can then be used to perform personalized profiling and recommendation for each user. In this paper, we introduce PathMining, an algorithm which constructs personalized user profile and make recommendations based on Monte Carlo sampling of graph traversal. PathMining predicts preferable items by emulating user's selection processes. The performance and potential value of our method is validated by using HetRec 2011 dataset.

Original languageEnglish
Title of host publicationProceedings of the 28th International Florida Artificial Intelligence Research Society Conference, FLAIRS 2015
EditorsWilliam Eberle, Ingrid Russell
PublisherAAAI Press
Pages519-522
Number of pages4
ISBN (Electronic)9781577357308
StatePublished - 2015
Event28th International Florida Artificial Intelligence Research Society Conference, FLAIRS 2015 - Hollywood, United States
Duration: May 18 2015May 20 2015

Publication series

NameProceedings of the 28th International Florida Artificial Intelligence Research Society Conference, FLAIRS 2015

Conference

Conference28th International Florida Artificial Intelligence Research Society Conference, FLAIRS 2015
Country/TerritoryUnited States
CityHollywood
Period05/18/1505/20/15

Fingerprint

Dive into the research topics of 'PathMining: A path-based user profiling algorithm for heterogeneous graph-based recommender systems'. Together they form a unique fingerprint.

Cite this