This hackathon is only open to students. Double check the event page for more information as this may mean only those from a particular university/country are eligible.
online
Event Type
6
Participants
0
Est. Projects
Favourite
Calendar
Passport
Share
Data fetched about 2 hours ago
About this hackathon
Uncover the Deceiver in a 48-Hour AI & NLP Kaggle Challenge!
About the Event
Step into The Liar's Chatroom, a fast-paced 48-hour online Kaggle competition where NLP meets digital forensics!
Participants will build an AI model to analyze fictional group-chat conversations, examine verified evidence, and identify which character is lying. Each conversation features four characters (A, B, C, and D), with exactly one character making a statement that contradicts the facts. Using a custom synthetic dataset, your mission is to train a model that spots contradictions with high accuracy.
️ Event Format & Dataset
Dataset: Participants receive a custom synthetic dataset containing labeled training data and unlabeled test data.
️ Approach: You may use any Machine Learning or NLP approach—including custom architectures, pretrained models, LLMs, fine-tuning, or rule-based heuristics.
Platform: Hosted externally on Kaggle.
Rules & Regulations
Team Size: Compete individually or in teams of up to 4 members.
Submission Limit: Maximum of 5 submissions per team per day on Kaggle.
Submission Format: Predictions must be uploaded in the prescribed CSV format before the deadline.
Fair Play: Manual labeling of the test dataset and sharing predictions/code between competing teams are strictly prohibited.
Code & Methodology Submission: Top-performing teams on the private leaderboard must submit a brief explanation/repository of their methodology and may be asked to present their approach to qualify for prizes.
Evaluation Criteria
Primary Metric: Evaluation is based strictly on Classification Accuracy on the dataset.
Final Ranking: Official winners are determined by the Private Kaggle Leaderboard following post-competition code verification.