基于自然语言理解的SPARQL本体查询

为了用户能够方便地获取本体知识，提出了基于自然语言理解的SPARQL本体查询。利用Stanford Parser分析用户的自然语言查询，根据语法构建查询三元组,与关键词的方法相比，有效地减少了组合的个数。结合用户词典，能较准确地把查询三元组的词汇映射到本体实体。分值计算时除了考虑词语的形式相似和语义相似外，还考虑了概念的模糊性，尽量返回具体的概念。利用本体推理获取隐藏在本体中的信息，对查询进行过滤和限制，提高了准确率。用户通过图形交互界面和系统进行交互，选择需要的结果，最后返回树形查询结果，并能看到相关的信息。实验结果表明，该方法达到了预期的效果。

SPARQL ontology query based on natural language understanding

For users can conveniently access to ontology knowledge, Simple Protocol And RDF Query Language （SPARQL） ontology query based on natural language understanding was put forward. Users natural language inquires were analyzed utilizing Stanford Parser, query triple was constructed according to the grammar, greatly reducing the number of combinations compared with the key word method. Combined with user dictionary, the terms of query triple could be more accurately mapped to the ontology entities. Scores calculation not only considered the similarity of words form and semantic, but also considered the ambiguity of concept, and returned the specific concept as far as possible. Employing ontology reasoning to obtain the information hidden in the ontology, the query was filted and limited to improve the accuracy. Users interacted with system through graphical user interface, selected the desired results, and finally returned query results in the form of tree, and related information could be seen. The experimental results show that the proposed method achieves the expected results.