The Design And Implementation Of An RDF Storage And Query System

The Semantic Web has emerged as the next generation of the World Wide Web, it aims at putting machine-understandable data on the Web where data can be shared and processed by automated tools as well as by people, this better enables computers and people to work in cooperation. In Semantic Web, RDF (Resource Description Framework) and RDF Schema are commonly used to describe metadata. With the widely application of RDF, the requirement of RDF storage and query increased greatly.This paper introduces the query languages in Semantic Web at first, especially the SPARQL query language, then describe the design and implementation of an RDF storage and query system (FIRTREE), FIRTREE uses forward chaining strategy, implements three functions, that is inserting data, deleting data and querying data. This paper also introduces a Semantic Web application-WonderSpace, this application shows an application scenario of RDF storage and query system.Many performance tests were conducted for current RDF storage and query systems, the results show that, in most cases, forward chaining based systems are superior to backward chaining based systems. However, when RDF data consists exclusively of a large class or property hierarchy that is both broad and deep, the disadvantage of larger storage space may at some point outweigh the advantage of faster querying. Based on the above considerations, we feel that the inference strategy employed by existing systems is not flexible enough for Semantic Web applications. Therefore, we design an RDF storage and query framework with flexible inference strategy, which can combine forward and backward chaining inference strategies. This paper also design a semi-automatic rule configuration tool, it is used for analyzing data characteristics, assists user by configuring the rules in inference engines.