论文部分内容阅读
认为处于多维社会网络中的用户会表现出多种行为取向和兴趣爱好,单独使用多维网络中的一维很难进行有效的社区发现。为了有效解决以上问题,首先基于用户关系紧密度将社交媒体中有向网转化为无向带权网,并将所有一维社交网络进行集成;然后利用SSN-LDA对社交用户进行隐含社区建模,以根据用户-隐含社区概率分布计算用户相似度;最后使用二分K均值进行用户社区划分。在真实科学网博客上进行试验,结果表明该方法能较好地进行用户社区划分。
Considering that users in multidimensional social networks may exhibit various behavioral orientations and interests, it is very difficult to find effective communities by using one dimension in multidimensional networks alone. In order to effectively solve the above problems, we firstly transform the directed net in social media into a non-directed one based on the user relational tightness and integrate all the one-dimensional social networks. Then SSN-LDA is used to implicitly construct social users Module to calculate the user similarity according to the user-implicit community probability distribution; and finally, the user communities are divided by two-point K-means. Experiments on the blog of Real Science Network show that this method can well divide the user community.