User song preferences using preference learning in artificial intelligence
Existing music industries are working on building high-precision music recommender system. The music recommender system must provide and efficient way to manage songs and help their customers to classify all the songs based on genres, artists, age groups, locations, and language by giving quality re...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Year Project / Dissertation / Thesis |
| Published: |
2022
|
| Subjects: | |
| Online Access: | http://eprints.utar.edu.my/4649/1/fyp_CS__2022_GBJ.pdf http://eprints.utar.edu.my/4649/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Existing music industries are working on building high-precision music recommender system. The music recommender system must provide and efficient way to manage songs and help their customers to classify all the songs based on genres, artists, age groups, locations, and language by giving quality recommendation. This is because users will experience difficulties choosing from the millions of songs. In this project, the main goal is to classify the millions of songs in accordance with the taste of the users. With a good recommender system, the user will spend less time to looking the songs and more time on enjoying their favourite song and discovering new favourite. Thus, a good quality music recommender system will have a strong user base and will create a strong thriving market. This project will implement various algorithms to compare the results with one another to figure out which is the most effective algorithm that is suited for a music recommender system. The most common model is popularity-based model because it is simple and intuitive. That another algorithm is collaborative filtering, this algorithm aims to find similarity between users, songs, and artists. This experiment will use SVD model, KNN model, based on metadata. The problems face when making this project is a metadata of song are too big and processing such a large dataset is GPU-intensive and require a lot of system memory. The accuracy results of the experiments very low due to large dataset. The collaborative filtering will be used in this project. The future work for this project is to provide enough storage for the song to be downloaded into the dataset so that the user can play it immediately rather than relying on YouTube links. |
|---|