Abstract
Every issue is treated differently, as some of them are resolved quickly, while others take longer to resolve. These issues are the major cause of frustration in users. Over time, it can also degrade the project’s health. To understand user behavior, we analyzed 748,791 stale issues from 1,736 projects. We defined four categories of stale issues based on their inactivity metrics. To analyze their thematic content, we used Latent Dirichlet Allocation (LDA) with sentiment analysis. Qualitative analysis was also performed on a sample of comments. Different stale issues’ categories have unique topics such as core architectural tasks, specific toolchain bugs, etc. Furthermore, we reveal a significant correlation between negative sentiment and technical pain points, especially build/dependency problems and persistent bug reports. This work contributes a fine-grained model of issue stagnation by identifying key markers of contributor friction, informing the design of tools to foster healthier collaborative environments.

This publication has 20 references indexed in Scilit: