Machine learning e redes sociais

: automatização da análise de comentários para estratégias de marketing

Abstract

The increased volume of data generated by users of social networks has required companies to adopt more efficient methods of analysis. Conventional systems have been shown to have limitations in the interpretation of textual content; therefore, there has been an adoption of Machine Learning (ML) techniques to process this data in an automated way. This study evaluated a system that automatically categorises customer comments on an insurance company's social media platforms. The system employed three ML algorithms: Naïve Bayes (NB), K-Nearest Neighbours (KNN) and Decision Trees (DT). The findings demonstrate that the NB algorithm attained 65,18% accuracy in sentiment classification, while KNN exhibited superior performance (85,06% accuracy) in differentiating between content generated by the company and by users. The categorisation of tags proved to be a more complex task, with NB algorithm reaching a hit rate of 55,28%. The results demonstrate the viability of applying ML techniques for the automated analysis of comments on social networks, providing companies with a valuable tool for monitoring customer perceptions and guiding more effective marketing strategies.

Date of Award	23 Jul 2025
Original language	Portuguese
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Vera Lúcia Miguéis Oliveira e Silva (Supervisor)