Towards Extracting Domain Knowledge from C Code

“Writing code is not the problem, understanding the code is the problem” this saying [7] summarizes how important is domain knowledge in software development and maintenance. Gathering this knowledge is an expensive process, which requires an investment of time, money, resources, and which is very demanding, because knowledge is scattered over various locations within source code. In this work, we propose a method to recover domain knowledge from C code using identifiers. To this end, we extract identifiers from source code, and we use them to generate domain concepts and investigate their interrelations. We describe four use cases, in which the domain concepts and their interrelations are typically used. To evaluate the performance of our approach, we conduct two experiments. Both experiments show promising result, in that our approach misses only few relevant concepts, and rarely generates irrelevant concepts. Currently, our approach is not fully automated, because the user has to traverse through a short list that contains both domain concepts and general concepts, and manually remove the general ones.