Computational analysis of hard and soft skills proficiency indicators based on usage data from the ChatSkills platform

• Grant number: 25/20991-9
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE
• Start date: July 01, 2026
• End date: March 31, 2027
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Principal Investigator: Eduardo Ramos Strucchi
• Grantee: Eduardo Ramos Strucchi
• Associated researchers: Francisco Elanio Bezerra ; LEILA MALTA HENRIQUE DA SILVA

Abstract

The convergence of educational technology, data science, and artificial intelligence has driven innovative methodologies for assessing competencies in digital contexts. The field of learning analytics has established itself as a central approach to understanding learning patterns, with significant advances in frameworks such as learning analytics for Learning and Growth Design and the European Framework for the Digital Competence of Educators, as well as in the use of psychometric and multimodal instruments. Recent research highlights the application of process mining, predictive algorithms, and multimodal analysis techniques, including natural language processing and computer vision, capable of mapping cognitive and socio-emotional competencies in a more contextualized way. Furthermore, studies indicate the potential of predictive dashboards and key performance indicators to support self-regulation and instructional intervention. Despite these advances, significant gaps remain. Many approaches still rely on subjective self-assessments or traditional quizzes, limiting the capture of nuances in soft skills. The systematic integration of granular behavioral data, such as textual interactions, response times, and engagement patterns, remains incipient, as does validation at scale, especially in Latin American contexts. This limitation compromises the practical applicability and scalability of competency inference models. Therefore, the ChatSkills platform represents a unique opportunity to advance in this scenario by offering a robust infrastructure for real-time educational data collection. Available data includes: student progress, course completion and non-completion, organizational affiliation, profile information, additional access data, quiz scores and the proportion of correct and incorrect answers, a competency map and index, and textual interactions with 315 artificial intelligence agents. The database also includes specific leadership metrics, access to profile maps, user interaction demands, metadata associated with eight competencies, and the generation of sub-competencies. Additionally, the platform records continuous activity logs and integrates over 700 structured sources via an Application Programming Interface. This project proposes to develop and validate a multimodal and explainable model for analyzing ChatSkills proficiency, combining learning analytics, process mining, and artificial intelligence. The expected contribution is the construction of a scalable, reliable, and adaptive system capable of analyzing proficiency indicators in hard and soft skills, supporting personalized learning paths and strategic decisions in professional training, with scientific relevance and practical impact. (AU)