面向电视观众的用户兴趣偏好建模方法

An Approach of Building User Interest Preference Model for TV Viewers

在电视节目的个性化推荐过程中,针对用户偏好难以显式获取的问题,本文提出了一种基于观看行为的用户兴趣偏好建模方法。该方法将用户观看电视节目过程中的点播操作和观看时长与电视节目的基本属性相结合,从时间和频率两个维度描述用户对电视节目的偏好程度,并建立用户兴趣偏好矩阵。在此基础上,针对2602个用户在7天内的116010条点播记录,采用top-N推荐算法进行实验。结果表明,该方法能够有效地描述用户兴趣偏好,提高电视节目推荐的准确率和召回率。

Aiming at the issue of hard to obtain user explicit preferences in Personalized TV program recommendation, an approach of building user interest preference model is proposed based on user viewing behavior. This approach not only uses the operation and duration in the process of viewing TV programs,but also combines with the basic properties of TV programs,then user interest preference is described from two dimensions of time and frequency, and the interest preference matrix is established. Our approach is measured by top-N algorithm on 116010 historical data of 2602 users watched TV programs during 7 days. The experimental results show that the proposed approach can describe user interest preference accurately and perform well in precision and recall.