Mejoramiento de la descripción de recursos educativos abiertos, a partir de técnicas basadas en inteligencia artificial, machine learning y minería de datos

Abstract

Although Open Educational Resources (OER) are fundamental for teaching, learning, and research thanks to open access policies and web tools their potential is often limited. Despite their exponential growth, many OERs are published with low-quality or incomplete metadata descriptions, which hinders their discovery, retrieval, and effective reuse in open digital repositories, leading to issues of ambiguity and inconsistency. To address this problem, a strategy is proposed based on machine learning techniques (Large Language Models - LLMs) and embedding techniques (vector representations) for semantic capture, aimed at improving the metadata elements that describe OERs. This strategy seeks to uncover new details that provide a better description of a resource, thereby maximizing the potential of OERs across various open digital repositories.