Relationship discovery and navigation in big graphs

Abstract

Relationship discovery is a challenging field, especially when handling with big graphs (tenths to hundreds vertices and edges). In this paper, we define a set of rules for relationship discovery. To evaluate them and find connections we implement these rules in Pregel Relationship Discovery (PRD) algorithm and also in our Graph Clutter Removal Tool (AGECRT). Our PRD algorithm is capable of navigation even through the opposite direction of edge, even without additional indexation. Graph visualization is important field in data overview creation. For visualization, we also propose a new edge coloring method based on hash codes obtained from vertices. When edges cross, not color but also a pen style can help to navigate edges. In this paper we join relationship discovery (represented by PRD) and graph visualization (with AGECRT). In order evaluate our rules and work we use PRD and AGECRT to present an experimental results from Freebase dataset.