Abstract
In this work, we consider the problem of combining link, content and temporalanalysis for community detection and prediction in evolving networks. Suchtemporal and content-rich networks occur in many real-life settings, such asbibliographic networks and question answering forums. Most of the work in theliterature (that uses both content and structure) deals with static snapshotsof networks, and they do not reflect the dynamic changes occurring overmultiple snapshots. Incorporating dynamic changes in the communities into theanalysis can also provide useful insights about the changes in the network suchas the migration of authors across communities. In this work, we proposeChimera, a shared factorization model that can simultaneously account for graphlinks, content, and temporal analysis. This approach works by extracting thelatent semantic structure of the network in multidimensional form, but in a waythat takes into account the temporal continuity of these embeddings. Such anapproach simplifies temporal analysis of the underlying network by using theembedding as a surrogate. A consequence of this simplification is that it isalso possible to use this temporal sequence of embeddings to predict futurecommunities. We present experimental results illustrating the effectiveness ofthe approach.