Graph-based Indexing Method for Searching in RDF Data

RDF is a generic graph-based data model of Semantic Web, and SPARQL is a query language for accessing the RDF data. With the increasing size of RDF data, answering complex SPARQL queries is expensive because multiple self-joins are needed to process. In this work, we consider an indexing and searching approach based on the graph structure of RDF data to reduce the number of join operations. It can speed up the queries’ performance and support chain and star shaped SPARQL query. Chain and star shaped subgraphs are extracted from the RDF data graph by considering the structure of edges around each vertex. The subgraphs obtained are stored as the index, named as CS-index. To execute a query, the query is firstly decomposed into query subgraphs based on the common join variable of its all triple patterns. And the query results are retrieved by searching the query subgraphs in CS-index, not in the whole data graph. The proposed index structure and searching approach tend to speed up the query response time by reducing the number of joins. We conduct a performance study on LUBM data set and see that our method outperforms the contest by a few orders of magnitude.