分布式RDF关键词近似搜索方法

发布时间：2018-03-24 01:38

本文选题：RDF　切入点：关键词　出处：《中国科学技术大学学报》2017年10期

【摘要】：现有的RDF关键词搜索方法主要是在大规模的RDF数据图上直接进行搜索,未能充分利用RDF本体中的语义信息,迭代次数过多造成搜索效率和效果不理想.针对这些问题,借助Redis内存数据库集群,提出分布式RDF关键词近似搜索算法(DKASR),即在分布式平台上实现大规模数据的并行搜索.算法结合RDF本体的语义信息构建本体子图,利用语义评分函数对本体子图进行排序,借助MapReduce计算模型实现并行搜索并返回Top-k结果;如果返回的结果没有达到Top-k,则对本体子图进行扩展生成近似本体子图,使用语义相似度函数对近似本体子图进行排序,再利用MapReduce计算模型实现并行搜索,直到返回Top-k结果.实验结果表明,DKASR算法能够高效正确地实现RDF关键词近似搜索并有效返回Top-k结果.
[Abstract]:The existing RDF keyword search methods are mainly based on the large-scale RDF data graph, and the semantic information in the RDF ontology is not fully utilized, so the search efficiency and effect are not satisfactory due to the excessive number of iterations. With the help of Redis memory database cluster, a distributed RDF keyword approximate search algorithm, named DKASRN, is proposed, that is, parallel search of large-scale data is realized on distributed platform. The algorithm combines semantic information of RDF ontology to construct ontology subgraph. The semantic scoring function is used to sort the ontology subgraph, and the MapReduce computing model is used to realize the parallel search and return the Top-k result. If the returned result does not reach Top-k, the ontology subgraph is extended to generate the approximate ontology subgraph. The semantic similarity function is used to sort the approximate ontology subgraph, and the MapReduce computing model is used to realize the parallel search. The experimental results show that the DKASR algorithm can efficiently and correctly realize the approximate search of RDF keywords and return the Top-k results effectively.
【作者单位】：福州大学数学与计算机科学学院;
【基金】：国家自然科学基金(61300104) 福建省科技拥军基金(JG2014001) 福建省自然科学基金(2012J01168) 福州大学科技发展基金(2013-XQ-32)资助
【分类号】：TP391.3

【相似文献】