Aplicación de técnicas no supervisadas para el análisis de textos relacionados a proyectos de innovación docente

Abstract

Abstract: The present degree work applies two techniques: Topic Modeling and Association Rules. Topic modeling, allows to identify a set of themes based on a corpus. As for the Association Rules technique, it is a technique that finds transactions. The objective of this paper is to analyze the teaching innovation projects, these are documents provided by the Laboratorio de Innovación e Investigación Docente of the Universidad Técnica Particular de Loja. Using Topic Modeling Better results were obtained by applying TF-IDF to the corpus, then bigrams and trigrams are created and finally the corpus is lemmatized. For the Association rules technique, the best results were reached when the terms were eliminated more frequently. The technique that gave the best results was Topic Modeling by identifying the main interests of teachers in terms of techniques and methodologies focused on the teaching-learning process. The results obtained from each technic are intended to provide the Academic Vice President of the UTPL with a general idea of the interests of teachers.