Bibliographic citations
Temoche, A., (2024). Comparación de algoritmos de machine learning para clasificar opiniones de usuarios acerca de publicaciones hechas por bancos de Perú sobre phishing [Universidad de Lima]. https://hdl.handle.net/20.500.12724/21106
Temoche, A., Comparación de algoritmos de machine learning para clasificar opiniones de usuarios acerca de publicaciones hechas por bancos de Perú sobre phishing []. PE: Universidad de Lima; 2024. https://hdl.handle.net/20.500.12724/21106
@misc{renati/1442587,
title = "Comparación de algoritmos de machine learning para clasificar opiniones de usuarios acerca de publicaciones hechas por bancos de Perú sobre phishing",
author = "Temoche Salinas, Adrian Alonso",
publisher = "Universidad de Lima",
year = "2024"
}
Phishing in recent years has increased and this year was no exception. Due to the pandemic, there was a considerable increase in this type of cyber-attacks; among the reasons for its growth is the quarantine. This type of attack usually has a negative impact on Internet users and even more so on companies, with phishing being one of the most frequent attacks. For example, in the news and on social networks it can be observed how banks or payment platforms warn their users about these attacks so that they can be alert, as can be observed the discomfort of users when they post or comment on social media that they have been victims of such attacks. For this reason, this work seeks to correctly identify people's opinions about publications made by banks in Peru about phishing, classifying their opinion as positive or negative. Natural language processing (PLN) and sentiment analysis are used to process the text, with the Naive Bayes, Random Forest and Support Vector Machine (SVM) algorithms in the case of machine learning and with the BERT transformer in the case of deep learning, in order to know which of the four algorithms has better results. The methodology used consists of 5 steps from the search and creation of the dataset, through the preprocessing stage, training, testing, and ending with the evaluation of the results. Of the four algorithms, BERT obtained the best result with an accuracy of 86.90% followed by Naive Bayes which obtained the best result in the cross validation with 79.62% accuracy and in the testing stage with 86% accuracy among the three machine learning algorithms. In addition, the results obtained show that the three machine learning algorithms have similar performance.
This item is licensed under a Creative Commons License