Publications

Design Pattern Detection (DPD) tools support the comprehension and maintenance of software systems, yet existing tools typically focus on a limited set of design patterns and programming languages. This study investigates the motivations behind these design decisions from the perspective of tool designers and explores how potential users perceive the usefulness of DPD tools in practice. Two online surveys were conducted involving DPD tool designers and software developers. The findings indicate that tool designers prioritize patterns with structures that facilitate automated detection and commonly target popular programming languages. From the users' perspective, DPD tools are valued for supporting program comprehension and software quality improvement, although usability issues, limited accuracy, and insufficient documentation remain major barriers to adoption. The results suggest that future DPD tools should better balance detection capabilities with usability to encourage broader practical adoption.

Este artigo apresenta uma investigação experimental conduzida durante um curso de Engenharia de Software (ES) na COPPE, Universidade Federal do Rio de Janeiro, com o objetivo de avaliar o desenvolvimento e o impacto de um Ambiente Virtual (VE) projetado para educação em ES, com foco específico em requisitos de software. O estudo avaliou como a natureza imersiva dos VEs contribui para otimizar os resultados da aprendizagem. Os resultados demonstraram que ambientes imersivos contendo elementos-chave de ES desempenham um papel crucial na melhoria da experiência educacional, melhorando o engajamento do aluno e a retenção de informações. Além disso, descobriu-se que o uso de VEs promove a colaboração e o desenvolvimento de habilidades práticas, ao mesmo tempo em que minimiza os riscos associados a erros do mundo real. Essas descobertas ressaltam o potencial dos VEs imersivos como ferramentas valiosas para o avanço de práticas educacionais modernas em ES.

The increasing capabilities of large language models (LLMs) offer a promising avenue for enhancing human-intensive software engineering activities, such as requirements elicitation. However, we lack strong empirical evidence on the impact of LLMs, such chatGPT, not only on the productivity and quality of requirement artifacts, such as user stories, but also as a requirements elicitation tool. To address this gap, this exploratory study investigates the impact of LLM used by 28 undergraduate students from two universities on documentation of user stories with use cases. Based on the Latin Square design, students created user stories with and without LLM support. We then analyzed the effects in terms of completion time, solution quality, students reported benefits and challenges, and interactions between LLM and students. As expected, our results indicate that participants using the LLM reduced the mean time to complete their tasks from 12 to 6 minutes. However, we observed major quality problems with the LLM-supported solutions, such as generic responses. Participants also reported other issues with using the LLM, such as becoming overly dependent on it and losing creativity. We identified a recurrent support-oriented interaction pattern between users and the LLM. Our findings suggest that educators and software professionals should exercise caution when using LLMs, as their use may negatively affect student learning and undermine a thorough understanding of specific software requirements.

The investigation of contributions from Latin American researchers is highly relevant in the context of the Metaverse for education and immersive learning. In this way, the objective of this study is to characterize the state of the art of Metaverse for Education based on Latin American research. Based on our findings, papers are predominantly focused on empirical and development studies, suggesting an under-representation of theoretical and conceptual approaches which are crucial for deeper understanding. Additionally, there's a significant opportunity to expand the research to other Latin American countries, besides Brazil, which would encourage a broader diversity of approaches and perspectives.

The Metaverse gained popularity with the rebranding of Meta, and since then, headsets, multimodal devices, and applications with immersive experiences have evolved and been applied in various domains such as marketing, education, and industry, among others. However, we question whether individuals with disabilities are also able to have immersive experiences. Therefore, this study aims to conduct a Rapid Review to identify the state of the art in accessibility within the Metaverse. From 145 studies identified in the Scopus search engine, 141 studies were excluded, resulting in 4 articles for data extraction. We found that there is significant ambiguity in the use of the term 'accessibility', which led to a high exclusion rate of studies. Furthermore, we did not identify any work that implements assistive technologies to enable immersive experiences for individuals with disabilities. For future research, we aim to conduct a more comprehensive systematic review to obtain more consistent results on the state of the art in accessibility within the Metaverse.

This work presents the development and usability evaluation of the Educational Metaverse Ecosystem (EME), which was designed to be collaborative and scalable, allowing the community to actively contribute virtual worlds. The evaluation, conducted with an expert, yielded scores of 57.5 for the Web Platform (good) and 37.5 for the Virtual Museum (poor). Despite these scores, the assessment helped identify critical issues and clear opportunities for improvement, reinforcing the pedagogical potential of the proposal. Furthermore, the results highlight that developing virtual worlds involves non-trivial technical and design challenges, requiring iterative and multidisciplinary approaches to ensure effective immersive experiences in educational contexts.