Implementing a Concept Analysis Tool for Identifying Abstract Data Types in C code

Many authors agree that programmer efforts are mostly devoted to maintaining software system This is mainly due to inappropriate documentation and degraded system-structure. To understand such systems, tools should be developed to facilitate the recovery of the system- architecture. We support the idea to extract the smallest components, that are significant at the architectural level, from source code to iteratively abstract larger components of the system-architecture. Abstract data types (ADT) have been reported to be among such atomic components. This thesis consists of four parts. At first, we give an introduction to concept analysis. Then, we analyse previous work on ADT-detection that is based on concept analysis.After that, we propose an improved approach to identify ADTs using concept analysis. Finally, we conduct a case-study on three medium-size C systems (30-38 kLOC) to compare the identified ADTs with the ADTs that were manually identified by software engineers.