Multi-modal Knowledge Graph Convolutional Networks for Music Recommender System

Emotion based Multimodal Music Recommendation System

Implements a state-of-the-art multimodal music recommendation system using the Multimodal Knowledge Graph Convolutional Network (MKGCN) architecture.

Features

• Multimodal Integration: Utilizes multiple data modalities to capture rich information about songs.
• Advanced Recommendation Algorithm: Employs the MKGCN architecture, which combines graph convolutional networks with multimodal embeddings to generate accurate and diverse recommendations.
• Interactive Interface: Provides a user-friendly web interface powered by PyQt5, allowing users to input their preferences and receive personalized song recommendations in real-time.

Sequence to run the files

1. For Multimodal-Knowledge Graph Convolutional Network

• Download this complete project and open in Visual Studio Code
• Connect to venv3.11 (Use Command: cd path_to_your_directory then venv3.11\Scripts\activate)
• Run Configure_Data.ipynb
• Utils.ipynb
• EMKGCN Data Loader.ipynb
• Multimodal_aggregator
• Neighbor_Aggregator
• Modal.ipynb (Make sure here you provide your system file path for (modal_0.tsv, modal_1.tsv....)
• EMKGCN Main (Change the Model Directory Path based on your system file path)

2. Connecting with Spotify API for most updated results

• Training_Spotipy
• Spotipy.ipynb (This is will playlist for 7 specified emotions)
• PyQt5.ipynb (Do use the paths specified from your system files) Executed!!

Required packages

The code 1 has been tested running under Python 3.6.5, with the following packages installed (along with their dependencies):

• pytorch == 2.0.0
• numpy == 1.14.5
• sklearn == 0.24.2