Electronic Thesis and Dissertation Repository

Degree

Doctor of Philosophy

Program

Electrical and Computer Engineering

Supervisor

Capretz, Luiz F.

2nd Supervisor

Ho, Danny

Co-Supervisor

Abstract

Lessons learned (LL) records constitute a software organization’s memory of successes and failures. LL are recorded within the organization repository for future reference to optimize planning, gain experience, and elevate market competitiveness. However, manually searching this repository is a daunting task, so it is often overlooked. This can lead to the repetition of previous mistakes and missing potential opportunities, which, in turn, can negatively affect the organization’s profitability and competitiveness. In this thesis, we present a novel solution that provides an automatic process to recall relevant LL and to push them to project managers. This substantially reduces the amount of time and effort required to manually search the unstructured LL repositories, and therefore, it encourages the utilization of LL. In this study, we exploit existing project artifacts to build the LL search queries on-the-fly, in order to bypass the tedious manual search process. While most of the current LL recall studies rely on case-based reasoning, they have some limitations including the need to reformat the LL repository, which is impractical, and the need for tight user involvement. This makes us the first to employ information retrieval (IR) to address the LL recall. An empirical study has been conducted to build the automatic LL recall solution and evaluate its effectiveness. In our study, we employ three of the most popular IR models to construct a solution that considers multiple classifier configurations. In addition, we have extended this study by examining the impact of the hybridization of LL classifiers on the classifiers’ performance. Furthermore, a real-world dataset of 212 LL records from 30 different software projects has been used for validation. Top-k and MAP, well-known accuracy metrics, have been used as well. The study results confirm the effectiveness of the automatic LL recall solution by a discerning accuracy of about 70%, which was increased to 74% in the case of hybridization. This eliminates the effort needed to manually search the LL repository, which positively encourages project managers to reuse the available LL knowledge – which in turn avoids old pitfalls and unleash hidden business opportunities.

Share

COinS