Towards mining replacement queries for hard-to-retrieve traces

摘要: Automated trace retrieval methods can significantly reduce the cost and effort needed to create maintain requirements traces. However, set of generated traces is generally quite imprecise must be manually evaluated by analysts. In applied settings when algorithm unable find relevant links for a given query, human user improve results adding additional search terms filtering out unhelpful ones. effectiveness this approach largely dependent upon knowledge user. paper we present an automated technique replacing original query with new terms. These are learned through seeding web-based then processing identify domain-specific The query-mining was fine-tuned using security regulations from USA government's Health Insurance Privacy Portability Act (HIPAA) traced against ten healthcare related specifications.