Aplicación de técnicas de procesamiento de lenguaje natural para la identificación de tópicos en textos provenientes de redes sociales

Abstract

Abstract: This work deals with the problem of insecurity in Ecuador, a topic that generates concern in society. In this study, data from the social network is used X (former Twitter), because citizens give their opinions on this topic. The goal The main objective of the research is to apply natural language processing techniques for the identification of topics related to insecurity. The methodology that was used consists of the use of two algorithms: LDA and BERTopic, for both models it was implemented the number of 8 and 15 topics to analyze them and see results in the identification of topics relevant. To evaluate the quality of the models, the coherence metric was used, which It allows determining whether the generated topics are consistent and coherent with each other. The Results showed that the most effective model was the LDA model, because this model presented clearer and more coherent topics, in which relevant topics stand out to violence, lack of security and the impact of crime on the daily lives of citizens.