Universidade Federal do Ceará
Mestrado e Doutorado em Ciências da Computação

Título: Adapting Returners and Explorers Dichotomy of Human Mobility to Understand Source Code Knowledge Distribution

Data: 29/08/2025
Horário: 14h
Local: Google Meet (https://meet.google.com/vaa-cdxz-bfe)

Resumo:

Understanding how source code knowledge is distributed among contributors is essential for maintaining software quality and project sustainability. Existing techniques, such as Truck Factor and developer ownership metrics, rely on contribution frequency and authorship but overlook developers’ behavioral patterns in navigating the codebase. In this paper, we adapt the Returners and Explorers (R&E) framework, originally proposed in human mobility studies, to characterize developer interaction with source code. Using the notions of “radius of code orbit” and its top-k variant, we classify developers as returners, explorers, or tourists based on their file change activity across 20 Apache open-source projects. Our results show that returners tend to specialize in specific code regions, while explorers exhibit broader engagement across the codebase. Despite being often overlooked, tourists form a substantial portion of contributors, especially in peripheral code files. We further compare R&E roles with expertise-based models like DOA, DOE, and Truck Factor, revealing that explorers are frequently aligned with expert classifications, whereas returners, despite their specialization, are sometimes underrepresented. The R&E framework complements existing models by capturing behavioral dimensions of knowledge, offering new insights into developer roles, collaboration patterns, and potential knowledge risks in large-scale software projects.

Banca examinadora: