中文摘要：

提出了一种用户兴趣扩展的方法以便应用于个性化推荐系统,对用户的搜索点击日志和浏览器的浏览日志进行统计,粗略对用户兴趣建模,从文本相似度、语言模型相关度、潜在的语义关联关系三个方面充分分析用户兴趣方向之间的关联关系,应用社区发现思想挖掘关联关系紧密的兴趣群组,并对用户兴趣在同一群组内进行适当扩展。通过试验结果分析,可以看出用户兴趣扩展对个性化推荐点击率的影响,并使点击率有近一倍的增长。

英文摘要：

An approach of user interest expansion was presented and applied into personal recommendation system, the basic idea was to make some statistics on user＇s browsing log and clicking log, the user＇s interest was roughly mod- elled. The associated relationship from the text similarity, the relevance of language model and potential semantic rela- tionship between the directions of user interest was analyzed, the interest groups using community detection method was identified, the user＇s interest was enriched appropriately in the same group. By experimental analysis, the impact of user＇s interest expansion on click rate in personalized recommendations was observed. The click rate had nearly doubled growth.