Computer Sciences and Information Technology
Title : | KauwaKaate*: A platform for fake-news verification (* inspired from the popular hindi song: jhoot bole kauwa kaate) |
Area of research : | Computer Sciences and Information Technology |
Focus area : | Development of a platform for fake-news verification |
Principal Investigator : | Dr Kameswari Chebrolu, Associate Professor, Indian Institute of Technology (IIT) Bombay |
Timeline Start Year : | 2019 |
Contact info : | chebrolu@iitb.ac.in ; sudarsha@cse.iitb.ac.in |
Details
Executive Summary : | The world is awash in fake news, thanks to the wide reach of social media platforms such as WhatsApp, Facebook, and Twitter. Fake news is alleged to have influenced the 2016 US election and has incited communal violence in India (e.g. the Basirhat incident). It even threatens public health (e.g. anti-vaccination and wrong medical advice). India suffers from fake news as much as any other country in the world. The goal of this project is to build an India-specific system that verifies news/claims by leveraging the latest state-of-the-art and building upon it. This system named KauwaKaate will provide an easy-to-use interface for the common man to fact-check any article of interest (including videos/images). This it does via WhatsApp integration, a web interface as well as a dedicated mobile app. The article submitted for fact-checking is then compared by our backend against a curated database. The source material for this database includes material from trusted fact-checking/news sites as well as sites that are well known for publishing fake or highly biased news. If no match is found, a new search is initiated by leveraging data mining, machine learning, natural language processing to gather data relevant to fact-checking the article. Based on this a true-false verdict is arrived at and conveyed to the user along with evidence for/against the claim; the long-term goal for this project is to automate the entire process of verification and generate human-readable evidence for/against a claim. However, the state of the art is not quite there, and humans need to be in the loop in most cases. For types of claims that cannot automatically be verified, this system will crowdsource fact-checking by scheduling tasks to people with the right expertise. The experts in turn are provided a user-friendly interface with relevant mined data as well as computational tools (video/image search, language translation services) to aid their decision making and search. Again, the decision is conveyed back to the user along with source-topic reliability or source-target bias, and stored in the database to directly answer future requests for the same claims/articles. The project team has an extensive background in Web application development, Web crawling, information retrieval, knowledge representation and database query processing. The current problem is an extension of these areas. In implementing the above vision, they will use the services of students and project staff to carry out relevant research and build the software system. They will divide the subtasks among the investigators to enable the work to be carried out in parallel, but with close interaction. Once this software is ready, they plan to use the reach of existing fact-checking sites (such as Check4Spam, whom we are in touch with) to publicize the tools and reach a wider audience, and also to help build a volunteer base who will help in manual fact-checking, making use of our crowdsourcing platform. |
Co-PI: | Dr S Sudarshan, Professor, Indian Institute of Technology (IIT) Bombay |
Total Budget (INR): | 74,53,600 |
Achievements : | • Abhishek Bagade, Ashwini Pale, Shreyans Sheth, Megha Agarwal, Soumen Chakrabarti, Kameswari Chebrolu, S. Sudarshan: The Kauwa-Kaate Fake News Detection System: Demo. COMAD/CODS 2020: 302-306
• Invited talk at Exploring Media Ecosystems Conference, March 2nd & 3rd, 2020, Samberg Conference Center at MIT. |
Organizations involved