Skip to main content

AI Innovation Talent Hunt 2025

Build effective ML recommendation systems through movie recommender and rating prediction challenges.

Competition Phases

Primary Phase: Movie Recommendation

Objective: Develop an automated system to recommend 5 movies for users who have reviewed at least 10 movies on IMDb.

Focus Areas:

  • Build personalized recommendations using historical review patterns
  • Leverage user preferences and behavioral data

Evaluation Metrics:

  • Recall@5, Recall@3, Recall@1
  • Any other metric of your choice

Bonus Phase: Rating Prediction

Objective: Develop an automated system to predict IMDb user ratings for movies based on movie attributes.

Focus Areas:

  • Train regression model using historical reviews
  • Predict how a user might rate an unseen movie

Evaluation Metrics:

  • Root Mean Squared Error (RMSE)
  • Mean Squared Error (MSE)
  • Any other metric of your choice

Important Dates

PhaseMilestoneDate
PrimaryCompetition StartsSeptember 23, 2025
Testing round StartsNovember 10, 2025
Final SubmissionNovember 20, 2025
BonusCompetition StartsNovember 22, 2025
Testing round StartsNovember 27, 2025
Final SubmissionNovember 30, 2025
ResultsResult Announcement and Prize Giving CeremonyDecember 4, 2025

Eligibility

Students

Must be current IUB students

Team Size

Individual or teams up to 3 members

Faculty Support

Up to 2 faculty supervisors (IUB only)

Prizes & Recognition

Cash Prizes

  • 1st Prize৳15,000
  • 2nd Prize৳10,000
  • 3rd/4th/5th Prize৳5,000 each

Additional Benefits

  • Online certificates for all valid submissions
  • Collaboration opportunities with MarriageChime and SimpliSolve LLC
  • Recognition at prize giving ceremony
  • Featured in future projects

Dataset

Primary Phase Dataset: Movie Data

Source: IMDb scraped data + IMDb Top 250

Attributes include:

  • Title, poster, release year, duration, IMDb rating
  • Genre, synopsis, user ratings, popularity score
  • Director, writer, cast
  • User reviews, critic feedback, awards
  • Metacritic score, filming location, production info

Validation Set: MovieLens Latest Small Dataset (included) serves as gold standard for evaluation.

Bonus Phase Dataset: User Review Data

Source: Scraped IMDb user reviews

Attributes include:

  • User ID, reviewed movie
  • User rating, review title & text
  • Spoiler flag, review date

Evaluation & Submission

Submission Requirements

  • Trained models + predictions (in specified format)
  • Evaluation on private test dataset to prevent overfitting
  • Submission instructions and format will be announced later

Judging Criteria

  • Model performance on private test set
  • Novelty and clarity of approach
  • Technical innovation and methodology

Important: Evaluation will be conducted on a private test dataset, so generalization is key. Participants must submit both predictions and trained models. The code must be reproducible on other computers. We recommend attaching a “README.txt” with the code, with clear instructions and descriptions.

Rules & Guidelines

Allowed

  • Any ML technique (collaborative filtering, deep learning, NLP)
  • Google Colab (Free/Pro), Kaggle Notebooks (Free GPUs for 30 hrs/week)
  • Local or personal setups
  • Pretrained models or external data (if properly cited and open-source)

Prohibited - Will Result in Disqualification

  • External proprietary data without disclosure
  • Plagiarism or model tampering
  • Violation of IMDb's Terms of Use
  • Overfitting to public test data

License & Ethical Use

  • Dataset shared for educational and research purposes only
  • Sourced from IMDb.com - must comply with IMDb's Terms of Use
  • Redistribution or commercial usage is strictly prohibited

Contact & Support

If you have any questions or need help:

Email: talenthunt@marriagechime.com

Competition Support: talenthunt@marriagechime.com

aiinnovationtalenthunt@gmail.com