Please use this identifier to cite or link to this item: http://hdl.handle.net/10609/132328
Title: Análisis de sentimiento de textos basado en opiniones de películas usando algoritmos de aprendizaje computacional
Author: Chulilla Alcalde, Jorge
Tutor: Isern, David  
Others: Ventura, Carles  
Abstract: Within the field of Natural Language Processing (NLP) an ecosystem of algorithms has been generated, allowing data analysis and classification using machine learning techniques. This work is focused on the so-called sentiment analysis, which allows to classify texts in the English language according to a polarity, in our case positive or negative. As a specific use case, we have decided on texts of opinions of movies or series using IMDB, considering the importance that the audiovisual sector has and the large amount of resources that this industry allocates for user preferences analysis. A standard three-month planning of tasks has been defined and followed, using a CRISP-DM type methodology, which is used for projects based on 'Machine Learning'. Within the initial analysis and after seeing the different techniques and algorithms dedicated to NLP, three different algorithms have been developed and tested, based on different concepts: firstly, the more classical algorithms, Multinomial Naïve Bayes and Logistic Regression, and finally ULMFiT, based on 'Transfer Learning' techniques. Finally, the results have been very good, with accuracies of around 90% in all three cases, being ULMFiT the best of all. In this sense, the resources necessary for this type of algorithm may not justify its use, considering that the differences has not been too big, but it does reflect its potential.
Keywords: sentiment analysis
machine learning
natural language processing
Document type: info:eu-repo/semantics/bachelorThesis
Issue Date: 27-Jun-2021
Publication license: http://creativecommons.org/licenses/by-nc-nd/3.0/es/  
Appears in Collections:Bachelor thesis, research projects, etc.

Files in This Item:
File Description SizeFormat 
jchulillaTFG0621memoria.pdfMemoria del TFG3,17 MBAdobe PDFThumbnail
View/Open
jchulillaTFG0621presentación.pdfPresentación del TFG2 MBAdobe PDFThumbnail
View/Open