Challenges for Domain Knowledge Driven Program Analysis.
Daniel Ratiu · 2009
Programming is a knowledge intensive activity. In or-der to write and evolve programs, programmers make use of knowledge of different kinds like about the business domain, programming technologies or design. Most of the current reverse engineering approaches do not explicitly take into consideration the domain knowledge implemented in pro-grams when the analyses are performed and thereby they lose important information. This situation is due to a series of difficulties that need to be overcome when performing analyses that interpret the code from the point of view of do-main knowledge. In this paper we advocate the need for the explicit use of domain knowledge in program analyses and we discuss the most important challenges that have to be addressed to realize such analyses in the practice: making the knowledge explicit, obtaining the knowledge, defining appropriate interpretations, and recovering the interpreta-tions automatically. We present islands of solutions to ap-proach these challenges and we argue that to systematically overcome them is required a community effort. 1