Capturing Users Requirements Using a Data Mining Approach

In success or failure of software project, software requirements plays a critical role. Requirement engineering is a human intensive process so the chances of errors, and failure to elicit and specify the right requirements, are more. The software practitioners have to be cautious about what is actually required by the stakeholders and elicit only those features and functionalities rather than some features that may look good but are impractical, and may lead to project failure or timeline delays, without adding much value to the project. In this paper, we introduced a data mining approach to capture user’s requirement for a generic software project, which in our case, was a campus management system (CMS). Initially, we started with a set of Software Requirement Specification (SRS) documents for similar type of projects done before and we extracted the most frequently appearing requirements which would potentially be the ‘Needs’ of the stakeholders, using text mining technique. After that, the low frequency requirements were extracted which could potentially be the ‘Wants’ of some particular stakeholder. Later on, we applied association rule mining to find out links between multiple requirements that led us to the requirement that frequently appeared together, as such links can be useful in identifying requirement dependencies and specification of further requirements. In order to validate the extracted requirements, elicitation sessions in the form of interviews and questionnaires were held with the potential stakeholders of the project and the results of the comparison between the extracted and elicited requirements are discussed at the end of this paper.