Creación de distractores para preguntas de opción múltiple mediante técnicas de incrustación (Word Embedding)

Abstract

The automatic generation of distractors for multiple-choice questions in academic documents relies on advanced word embedding techniques, such as Word2Vec. These techniques represent words as vectors in a semantic space, facilitating the discovery of relationships and similarities between them. Through these methods, plausible but incorrect distractors can be created that reflect common misconceptions or are semantically related to the question. This approach has enhanced the quality of assessments by providing more challenging and realistic answer options. Additionally, it has automated the process of creating quizzes, saving time for educators and ensuring greater consistency. However, it is necessary to monitor and adjust this process to avoid implausible incorrect answers or implicit biases. This innovative methodology is transforming the way assessments are designed and administered in educational settings and in academic documents such as guides, quizzes, and written evaluations. In this paper, we present an introduction to the topic, followed by an exhaustive literature review. Subsequently, we describe in detail the methods and algorithms employed. The results obtained from experimentation are analysed and discussed, culminating in robust conclusions and practical recommendations.