Manuscript Title:

A DEEP MULTIMODAL MODEL FOR FAKE TWEET DETECTION USING CONTENT BASED, METADATA AND DERIVED FEATURES

Author:

VAISHALI VAIBHAV HIRLEKAR, ARUN KUMAR

DOI Number:

DOI:10.17605/OSF.IO/QS3EZ

Published : 2022-12-23

About the author(s)

1. VAISHALI VAIBHAV HIRLEKAR - Sir Padmapat Singhania University Udaipur, Rajasthan, India.
2. ARUN KUMAR - Sir Padmapat Singhania University Udaipur, Rajasthan, India.

Full Text : PDF

Abstract

Social media has grown in importance over the last few decades because it allows people from all over the world to stay connected, however, it has become a breeding ground for misinformation during different public event as well as in the case of COVID-19 pandemic. Detection of fake information on social media has been technically challenging as it necessitates time-consuming evidence gathering and meticulous fact checking. There are generally three widely accepted characteristics of fake tweet: tweet content and associated features/metadata and credibility of the source. Using these characteristics, we propose a novel Fake Tweet Detection model in this paper that can identify whether given tweet is real or fake. The primary objective of the research is to identify fake tweets, and experimentation is carried out precisely utilizing COVID-19 fake tweets as a case study. We have used an ensemble model composed of tweet text features, metadata features and derived features present in the tweets. In order to assess the efficacy of the proposed methodology, we assessed our model using the COVID-19 dataset and achieved 99.42 F1-Score.


Keywords

Social media, Fake Tweet, Deep learning, Covid 19.