Collaborative Filtering with Users' Qualitative and Conditional Preferences

摘要

Current generation recommender systems are considered to be limited in: (1) utilizing users' qualitative choices and (2) tackling new items and new users. Our research focuses on building a new recommender system that utilizes users' qualitative and conditional preferences with Collaborative Filtering (CF). We call it CF with Conditional Preferences (CFCP). To represent users' conditional preferences in CFCP, we have developed Probabilistic TCP-net (PTCP-net). Intuitively, we argue that CFCP will be able to overcome the existing limitations, however we plan to do in-depth research on it.