Abstract

Poor requirements analysis process results in incomplete software applications. Some requirements appear as scattered and tangled concerns within requirements document. Hence it is difficult to identify such requirements. A number of research approaches such as Theme/Doc, early aspects identification, information retrieval and aspects identification using UML have been developed to identify crosscutting concern at the requirements level. Nevertheless, these approaches are only supported by semi-automated tools whereby human intervention is required to achieve the desired results. This research focuses on developing a tool to automatically identify crosscutting concern at the requirements level. A model based on Theme/Doc and early aspects identification approaches is formulated as the basis of this tool, 3CI. 3CI adopts natural language processing (NLP) techniques such as verb frequency analysis, part-of-speech tagging and dominant verb analysis. The tool usability, efficiency and scalability are evaluated by comparing the performance of a requirements engineer conducting similar task manually. Our evaluation on 3CI demonstrates 75% of accuracy.

Key words: Aspects-oriented requirements engineering, 3CI, crosscutting concern, dominant verb analysis.