Mapping Terms in Application and Implementation Domains

Abstract

One problem of reengineering is the gap between terms from the application domain and terms from the implementation domain. Particularly, it is observable while searching source code: in search queries maintainers often use terms from the application domain that differ from terms in the implementation domain that have been selected by original developers. Therefore, many queries fail. This paper proposes the utilization of code structure for automatic mapping application domain terms and implementation domain terms. In this way, implementation domain terms can be tagged by the identified terms from the application domain. Such redistribution of existing terms in the search index allows matching even those relevant documents which do not contain exact terms from the query. The feasibility of the proposed method was estimated by collecting statistics in an open-source project.