Experimental comparison of recommender systems
Recommender systems seek to predict the rating that a user would give an item, given the data of the past ratings of all users and items and other side information. Traditionally, recommender system methods are split into two broad categories: content-based and collaborative filtering approaches. Ho...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
2019
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/76935 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Recommender systems seek to predict the rating that a user would give an item, given the data of the past ratings of all users and items and other side information. Traditionally, recommender system methods are split into two broad categories: content-based and collaborative filtering approaches. However, because of the graph-structured nature of data in recommender system tasks, graph neural networks hold much promise in pushing the state-of-the-art in recommender systems. This project aims to investigate the latest graph neural network approaches to recommender systems. In particular, it aims to incorporate the use of Residual Gated Graph ConvNets, an architecture that has proven effective on various graph learning tasks, into the recommender system task. We show that within a framework similar to collaborative filtering, using graph neural networks can produce competitive results across various benchmark datasets. |
|---|