TRACKING HIVEMIND’S JOURNEYfrom inception to breakthroughs_

This paper explores the underexamined area of fine-grained emotion classification in app reviews, extending beyond the traditional focus on sentiment polarity (positive, negative, neutral). To capture the complexity of users’ affective responses, the study adapts Plutchik’s emotion taxonomy and introduces a structured annotation framework and dataset tailored to app reviews. Through an iterative human annotation process, the authors establish clear guidelines, highlight challenges in interpreting emotions, and assess the feasibility of automation with large language models (LLMs). The results show that LLMs substantially reduce manual annotation effort and achieve notable agreement with human annotators, though full automation remains difficult due to the nuanced nature of emotions. This work provides structured guidelines, an annotated dataset, and insights for building semi-automated pipelines, offering valuable contributions to opinion mining, requirements engineering, and user feedback analysis.

This paper investigates the potential of Multi-Agent Debate (MAD) strategies to enhance the performance of Large Language Model (LLM) agents in Requirements Engineering (RE) tasks. While prior research has focused on prompt engineering, fine-tuning, and retrieval-augmented generation, these methods often treat LLMs as isolated black boxes, relying on single-pass outputs with limited robustness and adaptability. Inspired by the way human debates improve accuracy by incorporating diverse perspectives, this study explores whether collaborative interactions among multiple LLM agents can yield similar benefits. We systematically analyze existing MAD strategies across different domains, identifying their key characteristics and developing a taxonomy of core attributes. Building on this foundation, we implement and evaluate a preliminary MAD-based framework for RE classification. The results demonstrate both the feasibility and potential advantages of applying MAD to RE, paving the way for more robust, adaptive, and accurate use of LLMs in engineering contexts.

This paper examines the current landscape of AI-powered software testing tools by systematically reviewing and classifying 56 commercially available solutions as of 2024. It analyses how these tools support different stages of the software testing process, ranging from test planning and test-case design to execution and maintenance, and highlights their potential to improve efficiency and effectiveness for test engineers. At the same time, the paper identifies key limitations, including false positives and insufficient contextual or domain understanding, which underscores the continued need for human oversight. The study argues that AI-assisted testing tools should be seen as complementary to human testers rather than fully autonomous solutions, with close human–AI collaboration remaining essential in the foreseeable future.