Sebis Research News

In large software engineering projects, designing software systems is a collaborative decision-making process where a group of architects and developers make design decisions on how to address design concerns by discussing alternative design solutions. For the decision-making process, involving appropriate individuals requires objectivity and awareness about their expertise. In this paper, we propose a novel expert recommendation system that identifies individuals who could be involved in tackling new design concerns in software engineering projects. The approach behind the proposed system addresses challenges such as identifying architectural skills, quantifying architectural expertise of architects and developers, and finally matching and recommending individuals with suitable expertise to discuss new design concerns. To validate our approach, quantitative evaluation of the recommendation system was performed using design decisions from four software engineering projects. The evaluation not only indicates that individuals with architectural expertise can be identified for design decision making but also provides quantitative evidence for the existence of personal experience bias during the decision-making process.

Abstract: The need to explicitly document design decisions has been emphasized both in research and in industry. To address design concerns, software architects and developers implicitly capture design decisions in tools such as issue management systems. These design decisions are not explicitly labeled and are not integrated with the architecture knowledge management tools. Automatically extracting these design decisions will aid the architectural knowledge management tools to learn from the past decisions and to guide architects while making decisions in similar context. In this paper, we propose a two-phase supervised machine learning based approach to, first, automatically detect design decisions from issues and second, to automatically classify the identified design decisions into different decision categories. We have manually analyzed and labeled more than 1,500 issues from two large open source repositories and have used this dataset for generating the machine learning models. We have made the dataset publicly available that will serve as a starting point for researchers to further reference and investigate the design decision detection and classification problem. Our evaluation shows that by using linear support vector machines, we can detect design decisions with 91.29% accuracy and classify them with an accuracy of 82.79%. This provides a quantitative basis for learning from past design decisions to support stakeholders in making better and informed design decisions.