Attributed Community Analysis: Global and Ego-centric Views

Xin Huang, Hong Cheng, Jeffrey Xu Yu

Research output: Contribution to journalJournal articlepeer-review


The proliferation of rich information available for real world entities and their relationships gives rise to a type of graph, namely attributed graph, where graph vertices are associated with a number of attributes. The set of an attribute can be formed by a series of keywords. In attributed graphs, it is practically useful to discover communities of densely connected components with homogeneous attribute values. In terms of different aspects, the community analysis tasks can be categorized into global network-wide and ego-centric personalized. The global network-wide community analysis considers the entire network, such that community detection, which is to find all communities in a network. On the other hand, the ego-centric personalized community analysis focuses on the local neighborhood subgraph of given query nodes, such that community search. Given a set of query nodes and attributes, community search in attributed graphs is to locally detect meaningful community containing query-related nodes in the online manner. In this work, we briefly survey several state-of-the-art community models based on various dense subgraphs, meanwhile also investigate social circles, that one special kind of communities are formed by friends in 1-hop neighborhood network for a particular user.
Original languageEnglish
Pages (from-to)29-40
Number of pages12
JournalIEEE Data Engineering Bulletin
Issue number3
Publication statusPublished - Sept 2016

Cite this