作者:Jie Tang
年份:2009
期刊:KDD
研究内容:区分不同angle(topic)上的社会影响,量化影响的大小。 propose Topical Affinity Propagation (TAP) to model the topic-level social influence on large networks.
¶主要思想
在主题层面(topic level)上利用亲和力传播(affinity propagation)来进行社会影响鉴定。
- TAP provides topical influence graphs that quantitatively measure the influence on a fine-grain level;
- The influence graphs from TAP can be used to support other applications such as finding representative nodes or constructing the influential subgraphs;
- An efficient distributed learning algorithm is developed for TAP based on the Map-Reduce framework in order to scale to real large networks.
¶数据
网络和节点下的主题分布。一个作者共现网络,一个引用网络,一个film-director-actor-writer网络
¶结论
相较于文本相似度有两个优势1.可以分析来年各个节点互相影响的差异。2.可以计算整个网络中影响力最强的节点。
借鉴意义:可对社交网络中用户节点转化成向量,然后计算用户节点与相连节点的相似性,作为亲密度度量的手段。