ReMo: A recommendation development model for software process improvement based on correlation analysis

초록

Continuous software process assessment and improvement are integral to the success of business objectives. Process assessment identifies the strengths and weaknesses in a software process and produces recommendations for planning improvements. The quality of recommendations is critical for constructive improvement planning and implementation. While widely practiced assessment models (e.g., CMMI, SPICE) address the identification of strengths and weaknesses and emphasize the importance of recommendations, they lack in providing concrete methods for developing quality recommendations. This leads to ad-hoc practices in building recommendations which often result in poor quality of recommendations as witnessed in a review of the current practice. To address this problem, we present ReMo, a novel recommendation development model that enables systematic development of quality recommendations through rigorous analysis of assessment findings. In ReMo, recommendations are developed through three phases: i) correlations analysis of findings and improvement package development, ii) review of improvement packages and refinement, and iii) recommendation development. ReMo is evaluated for its quality as a process and the quality of its outcomes through twelve industry case studies from various domains. The evaluation proves the effectiveness and usefulness of ReMo in building recommendations with room for improvement in its efficiency. The evaluation also witnesses the quality of resulting recommendations in terms of concreteness and comprehensiveness. © 2022 Scrivener Publishing.

키워드