A group of bachelor students from the HPI in Potsdam implemented a method to make exploration and understanding of complex graph structures easier. Tonight they want to explain the concepts behind it and show a demo of how this works on a few live graphs.
We introduce MetaExp, an open-source Neo4j-based graph exploration tool that can automatically harvest latent and implicit knowledge from large knowledge graphs. MetaExp incorporates the expertise of any user in the supervised definition of a unique and personalized exploration of the graph. At its core, MetaExp uses the expressive abstraction of meta-paths, which summarises walks over the graph. Empowered with such abstraction, the user can easily communicate their preferences and knowledge to the system. In the talk, we will showcase the system and the architecture, and we will share our experience on integrating complex analytical tasks into neo4j.