Summarizing RDF graphs using Node Importance and Query History

Abstract

RDF graphs have been widely used in various cognitive and intelligent services in industry. Recent tremendous growth in knowledge base data volumes has made searching and querying RDF graphs increasingly difficult. Summarization is an effective solution for managing large RDF graphs by extracting critical data into summary graphs. Improving the accuracy of the summary relative to the original graph and the relevance to actual user demands improves the efficiency and usefulness of the queries against the summary. In this paper, we present a hybrid summarization method that takes into account both the graph structure and user query history. Specifically, we define a hybrid metric of node importance that captures both the structural importance and user query preferences. We propose two algorithms to extract summaries of a given RDF graph based on this hybrid metric. We evaluate our approach in experiments using three public datasets (DBpedia, YAGO, and Freebase), and the results demonstrate the efficiency of our approach.