/SMAI_Project_30

Detecting rumour and stance jointly by "Neural Multi-Task Learning"

Primary LanguageJupyter Notebook

SMAI Final Project

Project Proposal

For the record

  • Project ID: 30
  • Team number: 7
  • Team name: Random_team_0
  • Project title: Detecting rumor and stance (bias) jointly by Neural Multi-Task Learning

Git reference:

Team Members

  • Animesh Das (20161302)
  • Ali Hussaini (20161204)
  • Thapaswini Kancharla (20161066)
  • Issac Balaji (20163051)

Main goal of the project

  • Implementing Neural Multi-Task Learning.
  • Detecting rumor and stance jointly using the above method.

Problem definition

  • Problem is to make a Neural Multi-Task Learning Model to detect rumor and stance jointly.
  • We'll first implement baselines for detecting rumours and bias. A good one to go with would be a logistic regression modle.
  • A further extension, which involves has been provided in the following paper: http://delivery.acm.org/10.1145/3190000/3188729/p585-ma.pdf
  • We choose to implement two out of the many models given in this paper.
  • A good place to start would be collecting data from Twitter as specified in section 6.1 of the above-mentioned paper.

Results of the project

  • Accuracy and confusion matrix.

Tasks for each member

  • Not yet decided.

Project milestones and expected time

  • Data collection:
    1. Rumor: 18-03-2019
    2. Stance: 20-03-2019
  • Model build: Not yet decided.
  • Testing phase: Not yet decided.

Presentation Link (Google Slides)