MSc Thesis Defense: Knowledge-Driven Framework for Team Formation in Open-Source Software Development by Najia Shinneeb

 

Abstract:

Team formation is a critical challenge in collaborative software development, as effective teams must bring together contributors with complementary skills, reliable participation, and the ability to collaborate cohesively. This challenge is particularly complex in open-source software projects, where teams are often decentralized, volunteer-driven, and characterized by uncertain contributor participation, varying expertise, and dynamic collaboration structures.

This thesis proposes a model that formulates open-source team formation as a multi-objective optimization problem, considering skill coverage, contributor reliability, long-term contributor participation, and social cohesion to improve team stability and project sustainability. To address these objectives, we employ a cultural algorithm-based optimization method to generate contributor teams under the identified constraints. We propose two cultural algorithm variants with different belief-space representations: Skill-CA, which captures knowledge based on contributor selection frequency, and Pair-CA, which captures knowledge based on pairwise collaboration frequency.

We evaluate our proposed framework using contributor information mined from real-world datasets and compare it with baseline methods to assess its effectiveness in forming stable, sustainable teams. Experimental results across four datasets show that Skill-CA generally performs better than the other models. Its advantage becomes more pronounced as the number of required skills increases, suggesting that incorporating extracted knowledge into the team formation process can substantially enhance its effectiveness.

Keywords: Team Formation Problem, Open-Source Software, Multi-Objective Optimization, Cultural Algorithms

Thesis Committee:

Internal Reader: Dr. Muhammad Asaduzzaman 

Internal Reader: Dr. Sherif Saad

Advisor: Dr. Pooya Moradian Zadeh

Chair: Dr. Saeed Samet