/IRE-major-Project---SemEval-Rumour-Detection

Primary LanguageJupyter Notebook

IRE-major-Project---SemEval-Rumour-Detection

Introduction

  • The first task involved Stance Classification of the responses to the source tweet is the first subtask for this project. This involves the categorization of these responses into the following categories:

    • Support: The author of the response supports the veracity of the rumor they are responding to.
    • Deny: The author of the response denies the veracity of the rumor they are responding to.
    • Query: The author of the response asks for additional evidence in relation to the veracity of the rumor they are responding to.
    • Comment : The author of the response makes their own comment without a clear contribution to assessing the veracity of the rumor they are responding to.

  • The second task of Veracity classification simply involved classifying the source post into “Rumour / Not a Rumour / Unverified” classes. For this task we decided to go with several baseline and deep learning based models. This was done to achieve maximum possible accuracy scores over a dataset which is significantly small in size.

Prerequisites

  • panda
  • numpy
  • tensorflow

Dataset used

https://figshare.com/articles/RumourEval_2019_data/8845580

How to run

  • Run this code tree2branches.py for preprocessing of data & convert it into proper formate.
  • Run all different .ipynb files for different model for Taks-A and Task-B on google colab

links