Protótipo de uma plataforma para gestão de projetos colaborativos interinstitucionais utilizando inteligência artificial

Abstract

Innovation management in multicampus public institutions faces barriers related to communication, information sharing, and the identification of competencies, associated with geographical dispersion and limitations in systems supporting collaborative projects. This dissertation aimed to develop and evaluate a technological artifact that integrates Natural Language Processing (NLP) techniques to support expert identification and the management of interinstitutional collaborative projects at the Federal Institute of Rio Grande do Sul (IFRS). The study was grounded in Design Science Research (DSR), structured around the Relevance, Rigor and Design cycles. In the Relevance Cycle, a questionnaire administered to 64 professionals identified communication between teams, data sharing, cultural differences, and expert allocation as the main barriers, supporting the definition of requirements organized into the dimensions of communication and collaboration, knowledge management, and decision support. The Rigor Cycle comprised the theoretical framework and bibliometric analysis, while the Design Cycle involved the development and evaluation of a Minimum Viable Product (MVP). The resulting artifact is characterized as a Technical-Technological Product (TTP), classified as Software/Application, designed to support managerial processes related to competency management, team formation, knowledge management, and decision-making in interinstitutional collaborative projects. The recommendation mechanism uses NLP, TF-IDF weighting, and cosine similarity to estimate the correspondence between project demands and researcher profiles, while optionally considering researcher availability. The evaluation with seven experts resulted in an average score of 91.1 on the System Usability Scale (SUS), classified as excellent, and six participants expressed their intention to use the solution. The results indicate that the artifact can support expert identification and project management by centralizing information and functionalities. However, the study assessed the perceived usefulness and usability of an MVP rather than organizational effects resulting from its prolonged implementation. [resumo fornecido pelo autor]