Searching Software Knowledge Graph With Question

Researchers have constructed a variety of knowledge repositories/bases in different domains. These knowledge repositories generally use graph database (Neo4j) to manage heterogeneous and widely related domain data, which providing structured query (i.e., Cypher) interfaces. However, it is time-consuming and labor-intensive to construct a structured query especially when the query is very complex or the scale of the knowledge graph is large. This paper presents a natural language question interface for software knowledge graph. It extracts meta-model of software knowledge repository, constructs question related Inference Sub-Graph, then automatically transfers natural language question to structured Cypher query and returns the corresponding answer. We carry out our experiments on two famous open source software projects, build their knowledge graphs and verify our approach can accurately answer almost all the questions on the corresponding knowledge graph.