| Autor: |
Zhuo, Xingrui1 zxr@mail.hfut.edu.cn, Qian, Shengsheng2 shengsheng.qian@nlpr.ia.ac.cn, Hu, Jun3 hujunxianligong@gmail.com, Dai, Fuxin4 fuxindai@tencent.com, Lin, Kangyi4 plancklin@tencent.com, Wu, Gongqing1 wugq@hfut.edu.cn |
| Zdroj: |
ACM Transactions on Information Systems. Nov2024, Vol. 42 Issue 6, p1-28. 28p. |
| Abstrakt: |
The article discusses advancements in Session-Based Recommendation (SBR) systems, particularly through a new model called the Multi-hop Multi-view Memory Transformer (M3T). Topics include the limitations of existing graph neural network (GNN)-based methods in handling feature ambiguity between sequence and item conversion information, the importance of accurately capturing user intentions through multi-view representations and the proposed k-order power method to manage item graphs. |
| Databáze: |
Library, Information Science & Technology Abstracts |
| Externí odkaz: |
|
