Researchon Storage And Query Of Ontology Based On NoSQL

Semantic web’s being proposed open the era that human and computer can understand each other.Thereby the network resources can carry a kind of metadata information, so that they can convey semantic informationamong the computers or between people and computer accurately.This kind of metadata is RDF(Resource Description Framework) data.With continuous development of the semantic web technology, RDF and OWL(Web Ontology Language) metadata are generally called ontology. While ontology technology is widely used in different areas, the scale of the ontology data is beyond the management ability of traditional storage systems. So,how to effectively storing and efficiently querying the massive ontology data become a difficult problem. The rapid development of the No SQL(Not only SQL) with distributed storage and computing technology provides a new solution for management of ontology data and a growing number of researchers of the semantic web have began to explore in this field.This thesis carried out works revolving around the No SQL and ontology storage and query as follows:①In the context of semantic web technology research, this thesis introduces the current development of related technologies and the existing ontology storage system based on traditional relational database, distributed system and Hadoop platform.②Summarized the advantages and disadvantages of the existing ontology storage model, this thesis proposes a ontology storage model based on HBase(Hadoop database).The OWL data are stored into two HTables separately according to the class and attribute and partitioned by column family,while RDF instance data are redundantly stored into three HTable tables respectively named SPO_C, POS_C and OSP_C, making full use of HBase’s rowkey with dictionary index and storing RDF data in these rowkeys.③Based on the ontology storage model proposed in this thesis, this thesis proposes triples matching algorithm, single triple pattern matching algorithm, the query algorithm in the BGP pattern, semantic query algorithm and query connection algorithm of graphs based on Map Reduce.A queue structure is used in the expansion of semantic query set, preventing the endless loop and supporting subclass, equivalence classes, sub-properties, equivalent properties, inverse properties and symmetric properties.④Under the distributed cluster environment,this thesis carries out experiments using the storage model and query algorithm designed in this thesis to test their performance, including load performance and query performance, on different LUBM(Lehigh University Benchmark) datasets.Compared with traditional relational database in data load test and contrastively query test on six kinds of LUBM data sets with different sizes also with comprehensive experiments and comparative analysis of different storage and query systems, this thesis verifies a superiority of the proposed ontology storage model and query strategy based on No SQL.