Entity and Relationship Labeling in Affiliation Networks

Many domains are best characterized as an affiliation network describing a set of actors and a set of events interlinked together in a variety of relationships. Relational classification in these domains requires the collective classification of both entities (actors and events) and relationships. We investigate the use of relational Markov networks (RMN) for relational classification in affiliation networks. In this paper, we introduce a novel dataset, Profile in Terror (PIT) knowledge base, that provides a rich source of various affiliation networks. We study two tasks, entity labeling and relationship labeling. We highlight several important issues concerning the effectiveness of relational classification. Our results show that the PIT dataset has a rich source of relational structure and therefore it is a useful dataset for statisical relational network learning community. 1