Introduction to algorithms
Computing communities in large networks using random walks
ISCIS'05 Proceedings of the 20th international conference on Computer and Information Sciences
Digging in the Digg Social News Website
IEEE Transactions on Multimedia
Generating graphs that approach a prescribed modularity
Computer Communications
Metric convergence in social network sampling
Proceedings of the 5th ACM workshop on HotPlanet
Hi-index | 0.00 |
As Online Social Networks (OSNs) become an intensive subject of research for example in computer science, networking, social sciences etc., a growing need for valid and useful datasets is present. The time taken to crawl the network is however introducing a bias which should be minimized. Usual ways of addressing this problem are sampling based on the nodes (users) ids in the network or crawling the network until one "feels" a sufficient amount of data has been obtained. In this paper we introduce a new way of directing the crawling procedure to selectively obtain communities of the network. Thus, a researcher is able to obtain those users belonging to the same community and rapidly begin with the evaluation. As all users involved in the same community are crawled first, the bias introduced by the time taken to crawl the network and the evolution of the network itself is less. Our presented technique is also detecting communities during runtime. We compare our method called Mutual Friend Crawling (MFC) to the standard methods Breadth First Search (BFS) and Depth First Search (DFS) and different community detection algorithms. The presented results are very promising as our method takes only linear runtime but is detecting equal structures as modularity based community detection algorithms.