基于语义和用户偏好的网络电视直播推荐方法

发布时间：2018-06-04 17:36

本文选题：网络电视直播 + 实时推荐　；参考：《微电子学与计算机》2016年12期

【摘要】：提出一种基于语义和用户偏好的网络电视直播实时推荐方法.该方法首先基于用户的历史记录构建用户偏好模型,然后使用基于词向量的语义相似度计算方法,分别计算待推荐节目和用户记录或待推荐节目和用户当前观看节目间的相似度,再结合该相似度和用户偏好求取用户对待推荐节目的虚拟兴趣,最后选出虚拟兴趣较高的一组节目作为对用户的实时推荐.实验结果表明,此方法的命中率在实时预测推荐的场景下较对比方法提高了10%以上,且在实时节目推荐的场景下有更好的推荐效果.
[Abstract]:A real-time broadcast real-time recommendation method based on semantic and user preferences is proposed. This method first constructs user preference model based on user history records, and then uses semantic similarity calculation method based on word vectors to calculate respectively the recommended programs and user records or the recommended programs and the user's current viewing programs. The similarity degree is combined with the similarity degree and the user preference to obtain the user's virtual interest in the recommended program. Finally, a group of programs with higher virtual interest is selected as the real-time recommendation for the user. The experimental results show that the hit rate of this method is 10% higher than the comparison method under the real-time prediction recommendation scene, and is recommended in the real-time program. There is a better recommendation in the scene.
【作者单位】：中国科学技术大学信息科学技术学院;上海文广互动电视有限公司;
【基金】：中科院先导课题(XDA060112030)
【分类号】：TP391.3

【相似文献】