Truth Tell Hackathon

Overview

In today’s fast-paced media environment, misinformation spreads rapidly, especially during live broadcasts. The challenge of detecting false information in real time is critical for broadcasters, journalists, and viewers alike. 

The TruthTell Hackathon aims to combat this problem by developing AI-powered tools capable of detecting, flagging, and providing real-time alerts on misinformation as it happens. These tools will empower broadcasters and viewers with accessible fact-checking during live events, improving transparency and helping the audience make informed decisions.

In this hackathon, participants will develop adaptable AI methodologies designed for dynamic media, focusing on real-time performance. Teams will explore advanced natural language processing (NLP) techniques, machine learning models, and integration with fact-checking databases to create innovative, scalable solutions.

Tech Stack

These are for your reference only. You are welcome to use any tech stack/language/database, etc.

Guidelines

• Teams will have access to relevant datasets, fact-checking APIs, and live broadcast feeds provided by the Ministry of Information (MIB) to aid in the development process. 
• Need to focus on AI  techniques like text classification,  entity recognition,  and knowledge graph integration to create real-time solutions.
• After prototyping and testing their tools against live data, teams will present their solutions to a panel of judges. 

Tasks

1. Dataset preparation
• Analyze the data provided by the Ministry of Information (MIB) and external fact-checking APIs
• Pre-process data, including cleaning, tokenization, and entity extraction for text-based media content
2. Developing a real-time NLP model
• Train machine learning or deep learning models on labeled misinformation datasets
• Implement NLP techniques such as text classification, sentiment analysis, entity recognition, and context extraction to flag potential misinformation
• Use transformer-based models for real-time text analysis of broadcast content
3. Fact-checking integration
• Integrate external fact-checking APIs into your solution for verifying flagged content
• Develop a system for cross-referencing live broadcast content with trusted knowledge databases
4. Real-time data processing
• Set up streaming infrastructure to process broadcast feeds in real-time
• Implement data pipelines that detect and process new information as it arrives
5. Knowledge graph integration
• Build and deploy a knowledge graph to track relationships between entities mentioned in the broadcast and their verified status
• Use this graph to assist in the detection of patterns indicative of misinformation
6. Real-time dashboard for broadcasters
• Create an interactive user interface that provides broadcasters and journalists with real-time alerts and fact-checks
• Design the dashboard to display alerts with confidence scores, sources, and verification information
7. Testing and validation
• Perform real-time testing with live or recorded broadcasts.
• Use ground-truth data from fact-checking organizations to evaluate the accuracy and reliability of your tool