Multi-label Genre Prediction

A genre prediction model that given a plot allows the user to predict possible genres for a movie or tv-show.

Deployment

The app can be found at: predict-genre.streamlit.app

Motivation

This model can be used to analyze synopses and identify potential mislabeling or inconsistencies in genre assignment. This can contribute to maintaining the quality and accuracy of genre metadata in media databases.

ML Pipeline

Underlying Model - BiLSTM

Given a huge description, Bidirectional LSTM (BiLSTM) with the ability to process input flows in both directions can significantly help with NLP and genre context development in our case

Nerual Architecture

• Input Neurons = 350 / 500
• Embedding Size = 300
• BiLSTM L1 Size = 128
• BiLSTM L2 Size = 64
• Dense Layer 1 Neurons = 64
• Dense Layer 2 Neurons = 27

Screenshots

Collaborators / Team Members

• Muhammad Abdullah
• Farheen Akmal

---