CAJ | 학술논문

传统的SPARQL查询引擎在处理查询时以三元组模式为基本单位做查询优化处理，在三元组模式较多时存在着过多的连接操作，开销比较大。文中基于文档数据库的存储和查询特点，提出一种利用主语分类的方式来存储RDF数据的方法，将不同的RDF三元组按主语分成不同的类，并存入文档数据库的文档中。在处理SPARQL查询时将三元组模式也按照主语分类，构成以主语相关块为单位的查询图，并提出一种基于属性相关性的选择度估计方法来优化查询执行计划。文中利用文档数据库CouchDB实现了新的SPARQL查询引擎，实验证明文中的方法能够提高SPARQL基本图模式查询的效率。
전통적SPARQL사순인경재처리사순시이삼원조모식위기본단위주사순우화처리，재삼원조모식교다시존재착과다적련접조작，개소비교대。문중기우문당수거고적존저화사순특점，제출일충이용주어분류적방식래존저RDF수거적방법，장불동적RDF삼원조안주어분성불동적류，병존입문당수거고적문당중。재처리SPARQL사순시장삼원조모식야안조주어분류，구성이주어상관괴위단위적사순도，병제출일충기우속성상관성적선택도고계방법래우화사순집행계화。문중이용문당수거고CouchDB실현료신적SPARQL사순인경，실험증명문중적방법능구제고SPARQL기본도모식사순적효솔。
Traditional SPARQL query engines optimize queries in terms of triple pattern as basic unit. This brings too many join operations in the face of SPARQL queries comprising many triple patterns,which lead to much query overhead. In this paper,taking the advantage of storing and quering of document-oriented database,propose a subject-classification approach to store RDF triples. RDF triples are parti-tioned into various classes in terms of identical subjects,and saved to the documents of the database. Triple patterns are classified accord-ing to their subjects as well,composing the query graph based on subject-related block. The method of selectivity estimation is improved on the new query graph. Using a document-oriented databse CouchDB to build a SPARQL query engine,the experiment proves the ap-proach is capable of improving the efficiency of SPARQL basic graph pattern query handling.