Unit:
Department of Informatics and TelecommunicationsΠληροφορική
Author:
CHANDRINOS THEODOROS-ALEXANDROS
ZAMPATIS THEODOROS
Supervisors info:
Δρ. Τσαλγατίδου Αφροδίτη, Αναπληρώτρια Καθηγήτρια, Τμήμα πληροφορικής και Τηλεπικοινωνιών, Εθνικό και Καποδιστριακό Πανεπιστήμιο Αθηνών
Original Title:
Τεχνικές Επεξεργασίας Φυσικής Γλώσσας για Εντοπισμό και Αποφυγή Ψευδών Ειδήσεων στα Μέσα Κοινωνικής Δικτύωσης
Translated title:
Natural Language Processing Techniques for Detecting and Avoiding Fake News on Social Media
Summary:
The advancement of social networks has facilitated the sharing and spread of news among people all over the world. With the growth of these networks and of the volume of news shared daily, the phenomena of fake news have become stronger and widely spread. Over the past few years, big social networks like Twitter admit that fake and duplicate accounts, fake news and fake likes exist in their networks. This stems from the fact that the social network account owners have the ability to distribute false information, to support or attack an idea or a product, to promote or demote an election candidate, as well as to influence real network users in their decision making. Therefore, misinformation detection in enhancing public trust and society stability becomes of critical importance. Along these lines, detection of misinformation is still a challenging problem for the Natural Language Processing community.
In our work, we have utilized natural language processing and supervised machine learning in order to detect fake tweets using Python. We have studied a variety of approaches on the subject from various sources and authors. This inspired us to combine these approaches with the goal to find out which combinations work better. Therefore, we have developed a software tool, which checks the success ratio of four (4) different systems for fake news detection using four (4) different datasets, resulting in a total of sixteen (16) ratios, one for each combination.
Main subject category:
Technology - Computer science
Keywords:
Fake News, Twitter, Algorithm, Machine Learning, Classifiers, Natural Language Processing, Vectorizers
