A Methodological Approach for Mining the User Requirements Using Apriori Algorithm

Abstract

Users of enterprise software are multiple, and their requirements are diverse. Often their specifications are masked by mundane details and at times are vague too. Acknowledging these complexities in requirements engineering, the paper proposes a multistage methodological approach based on Apriori algorithm, a data mining technique. It extracts useful information from the given data on the criteria of mutual association and sufficient frequency. The user requirements captured through interviews and brainstorming are pre-processed for eliminating unnecessary stop words and developing a uniform structure of small stories. Mutual association and occurrence of the requirements are represented through association rules and rule metrics, for example, ‘Lift', ‘Support', and ‘Confidence'. The requirements having strong and moderate association are placed in ‘Top Priority List'; those with nominal, weak, or nil association are placed in ‘Low Priority List'. Gap analysis is employed to validate the defined requirements with respect to stakeholders' expectations. The complete and correct lists of requirements significantly influence the client satisfaction, software development process, and its eventual success.