A systematic review of methods for business knowledge extraction from existing software systems

Abstract

Software maintenance and evolutions often result in large cost overruns and delayed delivery of required changes or improvements. As numerous studies have shown, adopting software to meet ever-changing business needs constitutes a major part of the software maintenance cost. The demand to facilitate software maintenance has led to the emergence of different methods for automated knowledge extraction from source code and other artefacts of existing software systems. This paper presents a systematic literature review of peer-reviewed conference and journal articles on this topic. The review has been undertaken to summarise the state-of-the-art in the research field, identify any gaps and explore possible directions for the further research. In this review, 7 digital libraries were searched and 24 papers dealing with the topic were identified and classified according to the four dimensions: extracted business knowledge kind, extraction techniques, kinds of software artefacts used as input sources, and extracted knowledge representation forms. The results of this study indicate that the research field is still immature and requires more comprehensive research. The results also show that there is a minority of methods that rely on widely adopted business knowledge classification schemes and only very few of methods employ standards for knowledge representation. It is believed that this review and classification scheme proposed in the paper would serve as a guide for both researches and practitioners in the further studies.