基于加权谱分析的用户网络社团协作推荐方法

A collaborative recommendation method based on user network community with weighted spectral analysis

推荐技术是解决信息过载的一种有效方法.为将纷杂的网络世界中人的行为和信息服务粘合在一起,提出了基于网络社团的协作推荐方法.利用加权谱分析提高特征向量的社团划分贡献度,充分考虑社团内用户的评价风格,将社团内用户的评价值依照用户评价偏好进行了均一化处理,最后按项目相似度对目标项目的评价进行预测.实验结果表明该方法具有较好的推荐性能.

Recommendation technology is an effective technology to alleviate the problem of ″information overload″. To agglutinate human＇s network behavior and network information service,a novel collaborative recommendation method based on network community was presented,which promoted the contribution of community division by weighting eigenvector of user network. As user＇s rating style is taken into consideration,an approach to normalize the user＇s rating values is proposed,and the rating based on item similarity within the same user community is predicted. The experimental results indicate that the method can achieve better recommendation performance.