Discovering Informative Subgraphs in Rdf Graphs

In most contemporary approaches to pattern discovery in graphs, either quantitative anomalies or frequency of substructure is used to measure the relevance of a pattern. In this thesis, we address the issue of discovering informative subgraphs within RDF graphs. In the context of Semantic Search, relevance of such subgraphs depends on the amount of useful information conveyed to a user. This in turn depends on the meaning (semantics) of the edges in the subgraph. We introduce heuristics that guide a discovery algorithm away from banal (both low information and low relevance) paths towards more informative and relevant ones. This guidance is based on weighting mechanisms (driven by relationships) for the edges in the RDF graph. We present an analysis of the quality of the generated subgraphs with respect to path ranking metrics. We then conclude by presenting intuitions about which of our weighting schemes and heuristics produce higher quality subgraphs.