Intelligent bug fixing with software bug knowledge graph

Abstract

Software bugs continuously emerge during the process of software evolution. With the increasing size and complexity of software, bug fixing becomes increasingly more difficult. Bug and commit data of open source projects, Q&A documents and other software resources contain a sea of bug knowledge which can be utilized to help developers understand and fix bugs. Existing work focuses on data mining from a certain software resource in isolation to assist in bug fixing, which may reduce the efficiency of bug fixing. How to obtain, organize and understand bug knowledge from multi-source software data is an urgent problem to be solved. In order to solve this problem, we utilize knowledge graph (KG) technology to explore the deep semantic and structural relationships in the multi-source software data, propose effective search and recommendation techniques based on the knowledge graph, and design a bug-fix knowledge question & answering system to assist developers in intelligent software bug fixing. At present, we have designed a bug knowledge graph construction framework, proposed the identification principles and methods for bug knowledge entities and relationships, constructed a preliminary knowledge graph based on the bug repository. In the following work, we will further improve the knowledge graph, complete the knowledge graph fusion of multi-source database, comprehend bug knowledge through knowledge reasoning, utilize the collaborative search and recommendation technology for bug-fixing knowledge question and answering.