Towards Community Detection in k-Partite k-Uniform Hypergraphs

Recently, numerous applications have emerged that create data most naturally in- terpreted ask-partitek-uniform hypergraphs. We identify benefits and challenges of generalizing community detection algorithms to these structures. We propose an algorithm which handles some of these challenges and a hypergraph generaliza- tion of the "caveman" model for the generation of synthetic evaluation datasets. The proposed algorithm outperforms a standard community detection algorithm working on a reduced, graph-like version of the original data, in particular when different domains possess differing community structures.