关联数据的自然语言查询方法-《计算机技术与发展》

文章信息/Info

Author(s):: XIAO Zheng; Department of Information Engineering,Sichuan Technology & Business College,Chengdu 611830,China

Keywords:: SPARQL query; metrics model; query semantic graph; natural language; associated data

摘要:: 以 RDF 结构为基础的数据网的发展中,高效数据检索成为关键问题之一。形式化查询语言( 如SPARQL) 因其语法的复杂性及查询本体的相关性阻碍其效用的发挥,迫切需要新的方法或工具实现以自然语言为基础( 如关键字检索) 的检索。形式化查询语言是检索这类结构化数据的有效方式,用户习惯自然语言为基础的检索方式。因而如何自动将关键词为基础的检索方式转换成以形式化查询为基础的检索方式是实现数据网的重要一环。关联数据的自然语言查询方法自动将自然语言查询转换成 SPARQL 查询,提高系统的有效性和效率。文中在抽象转换度量模型的基础上,以本体为基础构建查询语义图及实现语义消歧,构建 SPARQL 查询。实验结果表明,该方法具有更高的召回率、精度及更低的时间消耗。

Abstract:: Efficient data retrieval is one of the key issues in the development of data web based on RDF. Formal query language (such as SPARQL) are hampered by the complexity of their syntax and the relevance of the query ontology,so new methods or tools for natural language-based retrieval ( such as keyword retrieval) are urgently needed. Formal query language is an effective way to retrieve such structured data. Users are accustomed to natural language - based retrieval. Therefore, how to automatically convert keyword-based retrieval into formal query-based retrieval is an important part of the data network realization. The natural language query method of linked-data automatically converts natural language query into SPARQL query to improve its effectiveness and efficiency. On the basis of the abstract transformation metric model,the query semantic graph is constructed based on ontology and the semantic disambiguation is used to construct the SPARQL query. The experiment shows that the proposed method has higher recall and precision and lower time consumption.