Deseño e implementación dunha plataforma web para a análise e visualización de chíos sobre a COVID-19 en EEUU

Abstract

[Resumo]: O presente traballo de fin de grao centrouse no deseño e desenvolvemento dunha plataforma web dedicada á análise e visualización de mensaxes en Twitter (chíos) relacionadas coa COVID-19 nos Estados Unidos, basado na necesidade de comprender o impacto e a percepción pública da pandemia a través da recollida e análise de datos en redes sociais. O proceso estruturouse en varios pasos clave. Inicialmente, definíronse os requisitos funcionais e estableciuse o alcance do proxecto, para continuar coa obtención dun conxunto de datos adecuado ao problema. Seguidamente, procedeuse coa realización dunha análise preliminar dos datos dispoñibles, seguida dunha análise máis profunda que se centrou particularmente na detección e análise de sentimentos expresados nos chíos. Posteriormente, ideáronse bocetos para posibles interfaces de usuario que a plataforma podería adoptar, o que permitiu visualizar como os usuarios interactuarían coa aplicación, e desenvolveuse un modelado conceptual que serviu como base para a implementación. Finalmente, desenvolveuse e implementouse a plataforma. En canto ás tecnoloxías empregadas, elixíronse as seguintes: PostgreSQL, coa extensión PostGIS, empregouse como base de datos para almacenar o conxunto de datos recollidos. A extensión PostGIS permitiu almacenar, xestionar e analizar información xeo-referenciada de xeito sinxelo e integrado coa base de datos principal. O framework Flask, xunto co ORM SQLAlchemy (con extensión GeoAlchemy) foi a elección para definir os modelos de datos e crear a aplicación web. Para mellorar a interacción nas vistas finais, empregáronse tecnoloxías como Leaflet, HTML, CSS e JavaScript. Ademais, para analizar os sentimentos nos chíos, utilizáronse ferramentas de procesamento de linguaxe natural e análise de sentimento. O proxecto tamén fixo uso de PowerBI para a visualización efectiva do contido xerado. O traballo abarcou diversos aspectos esenciais para proporcionar unha ferramenta eficiente e significativa na comprensión da percepción pública durante a pandemia. Todo isto xestionouse seguindo unha metodoloxía iterativa e incremental, permitindo unha adaptación flexible ás necesidades en constante evolución e asegurando a calidade do resultado final ao longo do tempo.[Abstract]: The present final degree work focused on the design and development of a web platform dedicated to the analysis and visualization of Twitter messages (tweets) related to COVID-19 in the United States, based on the need to understand the impact and public perception of the pandemic through the collection and analysis of social media data. The process was structured in several key steps. Initially, the functional requirements were defined and the scope of the project was established, followed by obtaining a data set appropriate to the problem. Next, a preliminary analysis of the available data was performed, followed by a deeper analysis that focused particularly on the detection and analysis of sentiments expressed in the tweets. Subsequently, sketches were devised for possible user interfaces that the platform could adopt, which allowed visualizing how users would interact with the application, and a conceptual modeling was developed to serve as the basis for the implementation. Finally, the platform was developed and implemented. In terms of the technologies used, were chosen: PostgreSQL, with the PostGIS extension, was used as the database to store the collected dataset. The PostGIS extension allowed storing, managing and analyzing geo-referenced information in a simple and integrated way with the main database. The Flask framework, together with the SQLAlchemy ORM (with GeoAlchemy extension), was chosen to define the data models and create the web application. To improve interaction in the final views, technologies such as Leaflet, HTML, CSS and JavaScript were used. In addition, to analyze the sentiments in the tweets, natural language processing and sentiment analysis techniques were implemented. The project also made use of PowerBI for effective visualization of the generated content. The work covered various aspects essential to provide an efficient and meaningful tool in understanding public perception during the pandemic. All of this was managed following an iterative and incremental methodology, allowing for flexible adaptation to evolving needs and ensuring the quality of the end result over time.