一句话：提出负面情绪指数 NEI，基于 7.3 万条 issue 评论量化开发者对技术债的关注与挫败感。
技术债不仅影响代码质量，也持续影响维护它的开发者，但其情感成本缺乏可扩展度量。本文提出负面情绪指数（NEI），量化开发者对技术债相关关注点的情绪反应。作者分析 Apache Hadoop、Kafka 和 Flink 的 73,000 余条 issue tracker 评论，结合关键词债务挖掘、LDA 主题建模与 transformer 情绪分类，识别七类债务关注主题并比较其负面情绪强度。结果表明，运维与基础设施类关注（如不稳定的持续集成、配置复杂性和部署不稳定）比实现层和 API 可用性关注带来显著更高的负面情绪强度。文章还讨论将 NEI 集成到优先级框架和工程仪表盘，以支持更以人为本的技术债管理，并强调情感负担是可维护性研究中被忽视的维度。
Technical debt affects not only code quality but also the developers who must continuously manage its consequences. This paper introduces the Negative Emotion Index (NEI), a scalable metric for quantifying developers' emotional responses to technical debt-related concerns. Using over 73,000 issue tracker comments from three large Apache projects (Hadoop, Kafka, and Flink), we combine keyword-based debt mining, topic modeling, and transformer-based emotion classification to characterize frustration across seven latent debt-related concern themes discovered via LDA. We find that operational and infrastructure-oriented concerns—such as flaky continuous integration, configuration complexity, and deployment instability—are associated with systematically higher negative emotional intensity than implementation-level and API usability concerns. These results highlight emotional burden as an underexplored dimension of software maintainability, complementing established technical debt cost models. We further discuss how NEI can be integrated into prioritization frameworks and engineering dashboards to support more human-centered technical debt management.
@inproceedings{KassabSF26,
title = {The Emotional Cost of Technical Debt: Quantifying Developer Frustration Across Debt-Related Concern Themes},
author = {Mohamad Kassab and Izah Sohail and João Paulo Fernandes},
booktitle = {FSE},
year = {2026},
doi = {10.1145/3803437.3805542}
}