Research Development on Smart Recommendation Mechanism Enabled by Big Data in China
摘要: [目的/意义]旨在厘清国内大数据赋能的智慧推荐机制发展脉络与现状,凝练大数据赋能的智慧推荐机制。[方法/过程]文章用内容分析法归纳了521篇文献内容,从数据驱动的推荐机制改进、模型驱动的推荐机制优化、用户驱动的推荐结果混合三方面阐述了国内大数据赋能的智慧推荐机制研究进展。[结果/结论]大数据赋能的智慧推荐机制通过数据驱动改进数据源广度和特征提取深度,提升推荐输入环节特征工程质量;通过模型驱动优化、重构推荐模型,优化推荐处理环节模型匹配性能;通过用户驱动混合推荐结果,提升推荐表示环节服务效能。现有研究需强化数据驱动喂养式推荐、模型驱动可解释性推荐,创新用户驱动生成式推荐,构建数据、模型、用户三驱型智慧推荐机制体系。
Abstract: [Purpose/significance] In order to clarify the development and current situation of smart recommendation mechanism enabled by big data in China, and to condense the smart recommendation mechanism enabled by big data. [Method/process] Using the content analysis methods, the authors summary the content of the 521 articles, and expound the research development on smart recommendation mechanism enabled by big data in China from three aspects of improvement of data-driven recommendation mechanism, optimization of model-driven recommendation mechanism and mixing of users-driven recommendation results. [Result/conclusion] The smart recommendation mechanism empowered by big data improves the breadth of data sources and the depth of feature extraction through data-driven approaches, thereby enhancing the quality of feature engineering in the recommendation input stage. Through model driven optimization and reconstruction of recommendation models, model matching performance is optimized in the recommendation processing stage. Through user driven hybrid recommendation results, service efficiency is enhanced in the recommendation representation stage. Existing research needs to strengthen data-driven feeding recommendations and model driven interpretable recommendations, innovate user driven generative recommendations, and build a smart recommendation mechanism system driven by data, model, and user.
[V1] | 2024-09-30 20:33:43 | PSSXiv:202410.00152V1 | 下载全文 |
1. 数字化时代中国企业全球人才管理的法律合规与创新 | 2024-10-06 |
2. 基于知识图谱的深度问答及其应用研究 | 2024-09-30 |
3. 基于微服务架构的智慧档案馆知识服务架构体系研究 | 2024-09-30 |
4. 国内基于知识图谱的知识服务研究进展:核心内容 | 2024-09-30 |
5. 国内社交网络可视化研究进展 | 2024-09-30 |