Graph Clustering with Limited Marked Connections

Abstract

Cluster editing, also called graph clustering, of graphs with full information is a broadly researched topic. It has its roots in network science and computational biology. Closely related to this area is the research on graph clustering on graphs with missing information, or otherwise marked connections. This research aims to explore whether these marked connections can be used in a way to improve solutions of graph clustering. In this paper a new algorithm is proposed and tested against an existing algorithm for graphs with missing information. Their quality is compared, and the run-time of the existing algorithm will be improved upon with the new algorithm. In the end a new benchmark for algorithms concerning themselves with clustering graphs with marked connections will be given.