A Web Approach for the Extraction, Analysis, and Visualization of Sentiments in Social Networks Regarding the Public Opinion on Politicians in Ecuador Using Natural Language Processing and High-Performance Computing Tools

In a constantly evolving technological world, the presence of state-of-the-art computer-based methods such as Machine Learning (ML), Deep Learning (DL), and Natural Language Processing (NLP), among others has leveraged to develop systems for information extraction and sentiment analysis in raw text. Hence, the integration of these methods with the social network data is crucial to analyze posts, comments, or tweets on various subjects to capture social humor or sentiment around specific issues like the political arena as addressed in this paper. Hence, this research introduces a web application that enables interaction with web components and facilitates the extraction of posts and comments from Facebook related to political representatives using novelties technologies such as Selenium and ChatGPT. The latter has demonstrated an accuracy of 95%, a significantly high mark for a text processing-focused machine learning model. Hence, we have successfully categorized sentiments of social network data into positive, neutral, and negative. Also, we have found outstanding results by using an architectural parallel computing approach such as multiprocessing, where notable accelerations of 2.28× for extraction, and 2.69× for analysis have been achieved. As the digital environment continues to evolve, tools like these become fundamental for analyzing and comprehending social perceptions in an increasingly interconnected world. Therefore, the web application presented here represents a step forward in obtaining valuable information quickly and accurately, contributing to the understanding of public opinion regarding political and social matters, and not only limited to those issues but also to any subject addressed in society.