Prediction and Ranking Algorithms for Event-Based Network Data Joshua O'Madadhain, Jon Hutchins, and Padhraic Smyth   View Paper (PDF)   Return to December 2005, Volume 7, Issue 2 Abstract Event-based network data consists of sets of events over time, each of which may involve multiple entities. Exam- ples include email traffic, telephone calls, and research pub- lications (interpreted as co-authorship events). Traditional network analysis techniques, such as social network models, often aggregate the relational information from each event into a single static network. In contrast, in this paper we focus on the temporal nature of such data. In particular, we look at the problems of temporal link prediction and node ranking, and describe new methods that illustrate opportu- nities for data mining and machine learning techniques in this context. Experimental results are discussed for a large set of co-authorship events measured over multiple years, and a large corporate email data set spanning 21 months. ©2006 Association for Computing Machinery