Pattern Search in Temporal Social Networks

Due to the wide availability of social media and the wide range of real-life and human-centered applications, social networks have become an attractive research area. However, the temporal aspect of relations between entities in a social network has been widely ignored. We argue that the temporal aspect of social networks is the key to understand interactions and other phenomena happening in these networks and should thus be considered more closely. In this work we address the problem of pattern search in temporal social networks, thus finding all occurrences of a temporal pattern in a large temporal social network. As a first step, we define a temporal pruning criterion, which allows to quickly reduce the search space of candidates. Then, we present an index structure which allows to quickly find the occurrences of simple temporal network structures, from which more complex query structures can be derived from. Our experimental evaluation on a real-world temporal social network shows the effectiveness of our pruning approach and our proposed index structures, reducing the search-time for small temporal patterns by many orders of magnitude.