Linked Open Data Query Based on Natural Language

摘要

Linked open data (LOD) supports the SPARQL query strongly. A translation system from the natural language query to the SPARQL query based on the syntax rules is proposed. For a natural language query, a parsing method is proposed to represent the query intention and construct the corresponding query graph. The algorithms for obtaining and instantiating triple patterns are designed based on the rules. A mapping method for different types of graph nodes is lastly proposed to improve the recall. The experiments based on test data from QALD-4 are conducted. Compared with the other systems, our system is more easy and effective, and evaluation results are outstanding in the field of unsupervised learning.