Retrieving software maintenance history with topic models

Software maintenance history is a chronological record of significant maintenance events. It provides high level knowledge to software stakeholders, helps them to make important decisions. Unfortunately, retrieving software maintenance history is a heavy manual work so it can be invalid or even misleading. This paper use topic models to organize software maintenance documents such as bug reports into topics. Event detection tool is adopted to identify maintenance events from these topics. Finally, the identified events reconstruct software maintenance history. Retrieval results on 100,000 Eclipse bug reports are in accordance with Eclipse history records.