Design collaborative filtering recommender systems to solve cold-start problem
Recommender systems are information filtering system that suggests items like movies, songs, products, etc to users. Collaboration filtering approaches are adversely affected by the cold start problem, which makes it difficult to propose items to new users or for new items with no ratings when the i...
Saved in:
| 主要作者: | |
|---|---|
| 其他作者: | |
| 格式: | Final Year Project |
| 語言: | English |
| 出版: |
Nanyang Technological University
2022
|
| 主題: | |
| 在線閱讀: | https://hdl.handle.net/10356/156564 |
| 標簽: |
添加標簽
沒有標簽, 成為第一個標記此記錄!
|
| 機構: | Nanyang Technological University |
| 語言: | English |
| 總結: | Recommender systems are information filtering system that suggests items like movies, songs, products, etc to users. Collaboration filtering approaches are adversely affected by the cold start problem, which makes it difficult to propose items to new users or for new items with no ratings when the item is first launched or is never rated. As a result, the rating quality suffers. The sparsity of the rating matrix is also a significant issue, as it makes it difficult to identify items that are related to one another and are similar. Many techniques have been presented, all of which rely on asking users to manually rate various items. These techniques might not have optimum performance if the user is not interested or refuses to provide ratings manually.
The aim of this project is to provide improved memory and model-based algorithms to overcome the cold start problem faced by collaborative filtering algorithms and deliver better recommendations. The improved technique of this project entails populating the rating matrix to offer some ratings to new users in order to overcome the cold start problem while also decreasing the matrix's sparsity. The project results are compared to the results of probabilistic matrix factorization. |
|---|