基于关键字的RDF聚合查询研究  

RESEARCH ON AGGREGATE QUERY GENERATION BASED ON RDF KEYWORD SEARCH

作　　者：马晓芳[1] 杨卫东[1] Ma Xiaofang;Yang Weidong(School of Computer Science,Fudan University,Shanghai 201203,China)

机构地区：[1]复旦大学计算机科学与技术系,上海201203

出　　处：《计算机应用与软件》2023年第1期53-59,128,共8页Computer Applications and Software

基　　金：国家重点研发计划项目(2018YFB1004400)。

摘　　要：目前RDF数据上关键字查询转换为结构化语句的算法主要支持对于一般图元素的查询,而无法转换为包含聚合操作的结构化语句。关键字存在大量候选解释,且可能同时匹配聚合操作或图元素,这导致查询中聚合意图的理解非常困难。对此,提出将关键字查询自动转换为可能包含聚合操作的SPARQL语句的算法。算法对SPARQL所支持的聚合操作进行分类,获得关键字与聚合类别的匹配字典,进行关键字映射,计算关键字可能指示聚合意图的概率,确定候选查询解释,并利用模式图获得查询意图,设计意图分数计算方法和查询转换算法,得到对应的查询语句。LUBM和DBLP数据集上的实验验证了算法的有效性和准确性。While translating keywords query to SPARQL query on RDF, current algorithms can only support general graph queries, cannot query for structured statements containing aggregation operations. A large number of candidate explanations exist for keywords, and aggregation operations or graph elements may be matched at the same time, which makes it difficult to understand aggregation intentions in queries. An algorithm called PowerKTS is proposed, which can automatically convert keyword queries into SPARQL query statements, including statements with aggregation operations. Aggregation operations supported by SPARQL was classified, and the mapping dictionary of keywords and aggregation categories was obtained. The keywords were mapped. The query interpretation was determined by calculating the probability that the keyword may indicate the aggregation intent. The query intent could be acquired by expanding on schema graph. The score of aggregation was measured and a new assessment method of query intention was proposed to obtain SPARQL query statements. Experiments on LUBM and DBLP datasets verified the effectiveness and accuracy of the algorithm.

关 键 词：关键字查询 RDF SPARQL查询生成 聚合查询 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]