交互电视中基于本体的个性化节目协同推荐

An Ontology-based Personalized Program Collaborative Recommender for Interactive TV

提出一种在Web-TV环境中,拥有较强个性化和交互特性的基于本体的电视节目协同推荐方法。采用隐式和显式两种方法估计用户对其已收看节目的喜好程度,并根据用户收看电视节目的四条性质,提出隐式估计评分值的核心公式。在协同推荐时,利用节目本体中各元素的语义相似性,根据已经得到的评分值推测用户对未收看节目的评分值,解决了协同推荐的稀疏性缺点,并且在计算用户之间的相似度时,还考虑了用户的个人属性。最后还提出了实现了该算法的原型系统。

In this paper, a personalized and interactive collaborative recommendation method based on ontology in the Web- TV environment is presented. Specifically, we propose to estimate user preferences and ratings with both explicit and implicit information provided by users and raise a formula used to estimate ratings explicitly according to three properties. In collaborative filtering, we make use of the relations between the concepts of ontology and estimate the ratings of the programs that users didn＇t watch in terms of the program watched. Moreover, a prototype system is developed to illustrate the method in this paper.