摘要
针对网络电视的多媒体推荐系统,解决两个问题:数据源获取,以及优化推荐策略。数据源获取问题,通过用户的购买,播放等信息获得用户的兴趣度,即评分模型。优化推荐策略的问题,通过对用户进行聚类,将用户的数量级降低;用内容属性相似性关联和协同过滤推荐相结合的推荐策略,保证足够的推荐结果和高效的推荐质量。根据仿真结果,定性地分析了关键参数的意义。通过仿真和实际使用情况,说明该推荐系统是有效的。
Two problems are to be solved aiming at the multimedia recommendation system for internet television： the data source acquisi- tion, and the optimised recommendation strategy. For data source acquisition, users＇ interest degrees are gained through past information in regard to users purchase and play, i.e. the rating models. For optimised recommendation strategy, first, the order of users＇ magnitude is to reduce through users clustering; secondly, the recommendation strategy is employed to ensure enough recommendation results and efficient recommendation quality, the strategy combines the similarity association of content attributes with collaborative filtering recommendation. The significance of the key parameters are quantitatively analysed according to simulation results. The recommendation system is proved effective through the simulation and practical use.
出处
《计算机应用与软件》
CSCD
北大核心
2013年第1期79-82,共4页
Computer Applications and Software
基金
国家高技术研究发展计划(2011AA01A102)
国家科技支撑计划(2011BAH11B04)
中国科学院战略性先导科技专项(XDA06010302)
关键词
网络电视
推荐系统
数据源获取
协同过滤
内容关联
Internet television Recommendation system Data source acquisition Collaborative filtering Content association
