Hierarchical Aggregation Based Knowledge Graph Embedding for Multi-task Recommendation
Paper
Wang, Yani, Zhang, Ji, Zhou, Xiangmin and Zhang, Yang. 2023. "Hierarchical Aggregation Based Knowledge Graph Embedding for Multi-task Recommendation." 6th International Joint Conference on Asia-Pacific Web and Web-Age Information Management (APWeb-WAIM 2022). Nanjing, China 25 - 27 Nov 2022 Switzerland . https://doi.org/10.1007/978-3-031-25201-3_13
| Paper/Presentation Title | Hierarchical Aggregation Based Knowledge Graph Embedding for Multi-task Recommendation |
|---|---|
| Presentation Type | Paper |
| Authors | Wang, Yani, Zhang, Ji, Zhou, Xiangmin and Zhang, Yang |
| Journal or Proceedings Title | Proceedings of 6th International Joint Conference, APWeb-WAIM 2022 |
| Journal Citation | 13423, pp. 174-181 |
| Number of Pages | 8 |
| Year | 2023 |
| Place of Publication | Switzerland |
| ISBN | 9783031252006 |
| 9783031252013 | |
| Digital Object Identifier (DOI) | https://doi.org/10.1007/978-3-031-25201-3_13 |
| Web Address (URL) of Paper | https://link.springer.com/chapter/10.1007/978-3-031-25201-3_13 |
| Web Address (URL) of Conference Proceedings | https://link.springer.com/book/10.1007/978-3-031-25201-3 |
| Conference/Event | 6th International Joint Conference on Asia-Pacific Web and Web-Age Information Management (APWeb-WAIM 2022) |
| Event Details | 6th International Joint Conference on Asia-Pacific Web and Web-Age Information Management (APWeb-WAIM 2022) Delivery In person Event Date 25 to end of 27 Nov 2022 Event Location Nanjing, China |
| Abstract | Recently, knowledge graph has been used for alleviating the problems such as sparsity faced by the recommendation. Multi-task learning, which is an important emerged frontier research direction, helps complement the available information of different tasks and improves recommendation performance effectively. However, the existing multi-task methods ignore high-order information between entities. At the same time, the existing multi-hop neighbour aggregation methods suffer from the problem of over-smoothing. Also, the existing knowledge graph embedding methods in multi-task recommendation ignore the attribute triples in knowledge graph and recommendation tends to neglect the learning of user attributes. To mitigate these problems, we propose a multi-task recommendation model, called AHMKR. We use hierarchical aggregation and high-order propagation to alleviate the over-smoothing problem and obtain a better entity representation that integrates high-order information for multi-task recommendation. We leverage the text information of attribute triples, to improve the performance of knowledge graph in expanding the features of recommendation items. For users, we conduct fine-grained user learning based on the user attributes to capture user preferences in a more accurate matter. The experiments on the real-world datasets demonstrate the good performance of AHMKR. |
| Keywords | Graph neural network; Recommender systems; Multi-task learning; Knowledge graph |
| ANZSRC Field of Research 2020 | 460306. Image processing |
| Public Notes | Files associated with this item cannot be displayed due to copyright restrictions. |
| Series | Lecture Notes in Computer Science |
| Byline Affiliations | Nanjing University of Aeronautics and Astronautics, China |
| University of Southern Queensland | |
| Royal Melbourne Institute of Technology (RMIT) | |
| Zhejiang Lab, China |
Permalink -
https://research.usq.edu.au/item/z276z/hierarchical-aggregation-based-knowledge-graph-embedding-for-multi-task-recommendation
67
total views0
total downloads6
views this month0
downloads this month