Design and implementation of an intelligent requirements engineering tool for internet of things applications in an agile environment

Abstract

The decomposition of agile epics into user stories manually complicates sprint planning. If epics are poorly understood, they contribute to the threats regarding the sprint's completion. Performing the decomposition manual is laborious and complex and wastes resources in extensive projects. Natural language processing techniques present viable techniques that can automate the reduction of agile epics. This study explored and attempted to automate the decomposition of epics to their finest granularities, user stories and tasks using natural language processing (NLP). To decompose epics, we extracted and learned the essential parts of the linguistic structure of epics using NLP. The automation of agile epics refinement liberates the product owners from repetitive tasks and focuses more on managerial roles. The results of the decomposed epics were assigned to the task assignment model that uses the Hungarian algorithm to form sprints where team members were allocated tasks to attain a minimum time frame to complete the sprint. Furthermore, we then present our solution as a smart agile project management tool (SAPMT) that integrates the NLP techniques and Hungarian algorithm to assist project managers in the aspects of epic agile requirements decomposition and tasks assigned. The use of NLP has presented significant results in the generation of user stories and tasks from epics. The algorithm obtained an average accuracy of 89.25%, Precision of 100%, the recall of 77.25%, and the F1 Measure of 87%. The tool SAPMT was implemented using a python framework called Flask and presented a robust graphical user interface.