An Enhanced Recommendation Approach for Efficient Requirements Elicitation
Qusai Yousef Shambour, Nazem N. Qandeel, Hatem A. Shlool · 2024
The success of software projects depends on efficient requirements engineering, especially the crucial but difficult task of requirements elicitation. The excess amount of reusable requirements can cause information overload, hindering effective retrieval and possibly leading to overlooked or irrelevant selections. This paper presents an enhanced recommendation approach that integrates item-based collaborative filtering and content-based filtering methods to tackle this issue. Our approach uses stakeholder behavior patterns of known requirements and content similarity between requirements to help engineers manage information overload and select potential requirements that effectively meet their project objectives. Comprehensive evaluation using the RALIC and sparse datasets shows substantial enhancements in recommendation accuracy and coverage when compared to benchmark recommendation methods. Our proposed approach effectively addresses the data sparsity issue, a common challenge in recommendation systems. This study improves the efficiency and effectiveness of requirements elicitation, ultimately boosting the success of software development projects.