Please use this identifier to cite or link to this item:
Title: RumourOuT: rumours out
Authors: Mifsud, Philip
Keywords: Social media
Natural language processing (Computer science)
Application program interfaces (Computer software)
Rumor in mass media
Issue Date: 2018
Citation: Mifsud, Ph. (2018). RumourOuT : rumours out (Bachelor's dissertation).
Abstract: The circulation of rumours has become a world pandemic by the introduction of social media platforms, which bolster the spread of information in a global scale at relatively high speed. Social sites such as Twitter, is a catalyst of rumour diffusion which caused several outbreaks and confusion amongst its users. The proposed solution designed to tackle this problem is a rumour classification system called RumourOuT. It will make use of vital information that Twitter has to offer in terms of text data, user data and others that can motivate the development of such systems and improve the experience of the large community of users using this media platform. RumourOuT will exploit the data that Twitter has available to track and detect rumours with the aid of machine learning techniques. The developed solution contains several sections,subtask A applies sentiment classification on the tweets provided to view the user's thoughts on the main source tweet. The second subtask will decipher the final verdict of the rumour. These two subtasks will implement Natural Language Processing (NLP) techniques to extract information and metrics with the use of features to train the model that will lead to testing the same system. RumourOuT shows how it still performed efficiently in the classification task with a small unbalanced dataset, it obtained results that compete with other developed systems using the same dataset. Further experiments are also done to fully test the system and evaluate how it performs, with the introduction of a balanced dataset and a task in classifying replies concerning President Trumps' tweets.
Appears in Collections:Dissertations - FacICT - 2018
Dissertations - FacICTAI - 2018

Files in This Item:
File Description SizeFormat 
  Restricted Access
863.7 kBAdobe PDFView/Open Request a copy

Items in OAR@UM are protected by copyright, with all rights reserved, unless otherwise indicated.