中文摘要：

能耗社区是以用户为中心、绿色节能为主题的在线社区,它具有用户类别多,不同类别用户行为差别较大以及用户在线行为目的性强等特点.随着社区用户规模扩大,用户产生内容和社区资源不断增多,从而导致社区信息量过载.针对能耗社区的特点和其面临的问题,提出一个基于局部敏感哈希技术的能耗社区实时推荐系统.该推荐系统主要包括两个功能模块：1）离线数据处理模块,该模块采用局部敏感哈希技术对整个话题和资源集进行离线聚类处理;2）实时在线推荐模块,该模块及时获取用户在线行为和个人信息,为用户实时推荐感兴趣话题和资源.实验结果表明,本文提出的在线实时推荐系统能够在尽量降低数据处理量的情况下保证推荐质量,并根据用户当前在线行为,实时推荐用户感兴趣的内容.

英文摘要：

Energy-oriented online community is a user-centric and green energy-themed community. It reveals several characteristics such as multiple user categories,different user behaviors from different categories,strong purposes of online users and ect. With the number of community users growing,user-generated content and community resources increase rapidly,which results in information overload. Considering the characteristics and problems of this community,we propose a Locality Sensitive Hashing technology based real-time recommendation system for it. The proposed system consists of two function modules： 1） offline data processing module that adopts Locality Sensitive Hashing technology to classify topics and resources in the community offline; 2） real-time online recommended module that generates a recommended list for the target user in real time based on obtained user online behaviors and personal information. Our evaluation results reveal that the proposed system can minimize the amount of data processing,and rapidly recommend interesting contents in accordance with current online user behaviors in real time while guaranteeing recommendation quality.