NLBSE 2026

Abstract:

Natural Language Processing has become essential for mining insights from the textual artifacts produced and shared during software development—from issue reports and code reviews to developer discussions and app reviews. Over the past decade, the field has evolved dramatically: from lexicon-based approaches to machine learning classifiers to pre-trained transformers, and now to Large Language Models capable of zero-shot classification.

In this talk, I will reflect on lessons learned from building and evaluating NLP tools for software engineering. I will discuss recurring challenges—including domain adaptation, cross-platform generalization, and the critical role of data quality—that remain relevant regardless of the underlying technology. I will then examine what changes with LLMs: the opportunities they offer for addressing limited labeled data, but also new concerns around reproducibility, evaluation, and deployment. The talk will conclude with open challenges and directions for the NLBSE community.

Biography:

Nicole Novielli is an Associate Professor of Computer Science at the University of Bari, Italy. Her research lie at the intersection of software engineering and affective computing with a focus on understanding and supporting the human side of software development by investigating the role of emotional and cognitive states in software engineering. She investigates emotions through complementary approaches leveraging text-based analysis of developers' communication in software repositories, and physiological sensing using non-invasive biometrics. Nicole has co-authored more than 100 scientific papers. She is actively involved in the organization of scientific conferences, serving as Program Co-Chair of MSR 2022, SANER 2023, and ICPC 2026, and on the editorial boards of several scientific journals.

Abstract:

Open source software projects are complex socio-technical environments open for collaboration. However, contributors often have many questions about the project's codebase, processes, and community norms, and many communities lack the resources to answer them all. Much of the information required to answer these questions is already publicly available, scattered across software artifacts such as source code, documentation, issue trackers, commit histories, and comments. In this talk, I present our group's research aimed at helping developers find answers to their questions in natural language, including the development of a chatbot powered by an RAG approach using large language models to provide contextual answers grounded in project artifacts and a tool that connects questions to metrics derived from repository mining techniques. I discuss the lessons learned, the challenges encountered, and the implications of these approaches.

Biography:

Marco A. Gerosa is a full professor in the School of Informatics, Computing, and Cyber Systems at Northern Arizona University and is also affiliated with the graduate program at the University of São Paulo (Brazil). His research output spans broadly in empirical software engineering, with particular emphasis on open-source software (OSS) ecosystems, mining software repositories (MSR), and human aspects of software development. Gerosa's research interests include the intersection of artificial intelligence and software engineering, exploring how conversational agents and large language models can support developer tasks. More recently, his research investigates how software engineering education and OSS contribution by students can be supported through gamification and new AI-driven tooling.