Interactive Recommendations in Probabilistic Network on Ontology-based Preferences with User Feedbacks

摘要

Recommendation systems have become increasingly popular in recent years by providing users with personalized content and services. Collaborative filtering is a technique to make automatic predictions by collecting personal information from many users. Such systems are effective to deal with extracting useful information of users in a large dataset, but they are not aware of particular user intentions for the interest-based recommendations. In this paper, we show an interactive recommendations in probabilistic network with user-friendly questions from Ontologies that contains the background of user information. We also maintain belief networks to reflect user preferences and take feedback data repeatedly to consider user satisfaction.