Empirical Analysis of Knowledge Representation for Anime Recommendation Using Graph Neural Networks

被引:0
作者
Saito, Yuki [1 ]
Egami, Shusaku [2 ]
Sei, Yuichi [1 ]
Tahara, Yasuyuki [1 ]
Ohsuga, Akihiko [1 ]
机构
[1] Department of Informatics, The University of Electro-Communications
[2] Artificial Intelligence Research Center, National Institute of Advanced Industrial Science and Technology
关键词
data integration; graph neural networks; knowledge graph; recommender system;
D O I
10.1527/tjsai.39-6_AG24-D
中图分类号
学科分类号
摘要
In recent years, entertainment content, such as movies, music, and anime, has been gaining attention due to the stay-at-home demand caused by the expansion of COVID-19. In the content domain, research in the field of knowledge representation is primarily concerned with accurately describing metadata. Therefore, different knowledge representations are required for applications in downstream tasks. In this study, we aim to clarify effective knowledge representation for predicting users’ latent preferences through a case study of an anime recommendation task. We developed hypotheses from both quantitative and qualitative aspects on how to represent work knowledge to improve recommendation performance, and verified them by changing the structure of knowledge representation according to the hypothesis. Initially, we constructed a Knowledge Graph (KG) by integrating domain-specific and general-purpose data sources through the process of entity matching and imposing constraints on the properties. Subsequently, we constructed multiple KGs by varying the knowledge configuration. Specifically, we changed the composition of the data sources considered in the KG construction or excluded a triplet associated with an arbitrary property. After that, we fed the constructed KGs into the graph neural network recommender model and compared the recommendation performance. As a result, it was shown that the recommendation performance based on the KG composed of multiple data sources was the best, thus supporting the hypothesis from a quantitative aspect. Next, an ablation study on the properties revealed that knowledge characterizing the work itself contributed to the recommendation performance, thus supporting the hypothesis from a qualitative aspect. Furthermore, we constructed a text-based KG by generating a new vocabulary from the “synopsis” text. It can describe the work’s storyline and worldview in more detail. We take it as an input to a Large Language Model (LLM) and extend the existing metadatabased KG. The results showed that the KG considering both metadata and text had the best overall recommendation performance, again confirming the hypothesis. © 2024, Japanese Society for Artificial Intelligence. All rights reserved.
引用
收藏
相关论文
共 39 条
[1]  
Auer S., Bizer C., Kobilarov G., Lehmann J., Cyganiak R., Ives Z., DBpedia: A nucleus for a web of open data, International Semantic Web Conference, pp. 722-735, (2007)
[2]  
Bollacker K., Evans C., Paritosh P., Sturge T., Taylor J., Freebase: a collaboratively created graph database for structuring human knowledge, Proceedings of the 2008 ACM SIGMOD International Conference on Management of Data, pp. 1247-1250, (2008)
[3]  
Brown T., Mann B., Ryder N., Subbiah M., Kaplan J. D., Dhariwal P., Neelakantan A., Shyam P., Sastry G., Askell A., Et al., Language models are few-shot learners, Advances in Neural Information Processing Systems, 33, pp. 1877-1901, (2020)
[4]  
Carothers G., Seaborne A., Beckett D., RDF 1.1 N-Triples, W3C Recommendation, (2014)
[5]  
Chia Y. K., Bing L., Poria S., Si L., RelationPrompt: Leveraging prompts to generate synthetic data for zero-shot relation triplet extraction, Findings of the Association for Computational Linguistics: ACL 2022, pp. 45-57, (2022)
[6]  
Consumer Affairs Agency 21] Consumer Affairs Agency 消費者庁, (2021)
[7]  
Cyganiak R., Wood D., Lanthaler M., Klyne G., Carroll J. J., McBride B., RDF 1.1 concepts and abstract syntax, W3C recommendation, (2014)
[8]  
Du Y., Zhu X., Chen L., Zheng B., Gao Y., HAKG: Hierarchy-aware knowledge gated network for recommendation, Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval, pp. 1390-1400, (2022)
[9]  
Gilmer J., Schoenholz S. S., Riley P. F., Vinyals O., Dahl G. E., Neural message passing for quantum chemistry, Proceedings of the 34th International Conference on Machine Learning, Vol. 70 of Proceedings of Machine Learning Research, pp. 1263-1272, (2017)
[10]  
Goldberg D., Nichols D., Oki B. M., Terry D., Using collaborative filtering to weave an information tapestry, Commun. ACM, 35, 12, pp. 61-70, (1992)