Desarrollo de un prototipo de sistema para el análisis de opiniones basado en tweets. Caso de estudio: metro de Quito.

Abstract

Twitter is a social media platform designed to share immediate information about various topics, including political, social, and national issues. This project aimed to analyze opinions on Quito's subway through the CRISP-DM methodology, which involved collecting tweets and processing the data to ensure information quality. The GSDMM model was then used for topic generation and identification of dominant topics, and a supervised learning model, SVM, was trained with Spanish data for sentiment analysis to determine tweet polarity. A dashboard was developed as a prototype for web development using Flask, Dash, and Python as the programming language. Pandas was used to structure the data, and Plotly was used to create graphics to be displayed on the dashboard. The analysis uses visual storytelling to easily display the most relevant topics, such as progress, budget, subway administration, and construction issues, to identify user viewpoints and polarity of feelings (positive and negative). The dashboard is divided according to the different periods of Quito's mayoralties between 2015 and 2023.