Please use this identifier to cite or link to this item:
Title: Toxicom: Detección de mensajes tóxicos en medios sociales
Author: Torné Alonso, Raúl
Director: Casas Roma, Jordi
Tutor: Valdivia García, Ana
Others: Universitat Oberta de Catalunya
Keywords: artificial intelligence
natural language processing
toxic comments
Issue Date: 3-Jun-2018
Publisher: Universitat Oberta de Catalunya
Abstract: The purpose of this project is to automatically detect those messages whose purpose is negative in the social media. For this, Artificial Intelligence techniques based on Machine Learning and Deep Learning algorithms will be used. It should be borne in mind that there is currently a great diversification of social media, in which a large part of the population is continuously active. Thus, some of these important as Twitter, Facebook and Instagram among many others, generate millions of publications daily. So, it is necessary that those social media in which users generate content, will be needed controlled in this way, for avoid messages with negative intent and that at the same time generate hate. However, processing the manual detection of these messages would take a large amount of time. For this reason, most platforms currently integrate a request for manual prohibition. Although it must be said that only those that have been banned are evaluated, a problem that generates a great delay and why many users may have suffered damages when the ban is applied. At this time, to avoid the large amount of analysis and processing time of each of the generated messages, it is necessary to implement new models that allow the messages to be classified according to their content automatically. In this way, it will allow the manager of the social environment to indicate which types of messages will be accepted or not be published.
Language: Spanish
Appears in Collections:Bachelor thesis, research projects, etc.

Files in This Item:
File Description SizeFormat 
rtorneal10TFM0618memoria.pdfMemoria del TFM2.83 MBAdobe PDFView/Open

Items in repository are protected by copyright, with all rights reserved, unless otherwise indicated.