| 基于用户多模态输入提升推荐效率的商品推荐系统研究 |
| 点此下载全文 |
| 引用本文：王新成,许壮,郑健.基于用户多模态输入提升推荐效率的商品推荐系统研究[J].上海第二工业大学(中文版),2025,42(3):316-322 |
| 摘要点击次数: 406 |
| 全文下载次数: 60 |
|
|
| 中文摘要:为进一步提高推荐系统的效率, 从用户多模态输入角度出发, 提出融合文本、图像、音频等多种方式构建用户兴趣特征, 并利用不同模态间的互补关系建立用户多模态特征。通过文本和图像等多模态数据对用户行为进行建模, 并对用户多模态特征进行融合, 从而提升推荐系统的整体性能。实验结果表明, 基于深度学习算法的系统推荐耗时22.85 s, 而多模态模型仅需0.124 s, 相对于深度学习算法, 可以节省99.47% 的时间, 显著提高了商品推荐速度。此外, 多模态模型的推荐准确率始终优于深度学习算法, 当迭代次数为100 时, 其准确率为89.5%, 较深度学习算法提升13.86%。 |
| 中文关键词:多模态模型 推荐效率 推荐系统 内存 准确率 |
| |
| Research on Product Recommendation System Based on User Multimodal Input to Improve Recommendation Efficiency |
|
|
| Abstract:In order to further improve the efficiency of recommendation systems, it is proposed to from the perspective of user multimodal input, it proposes to integrate multiple modalities such as text, images, and audio for constructing user interest features, and constructs user multimodal features through complementary relationships between different modalities. Modeling user behavior through the use of multimodal date such as text and images is constructed, and multi-modal features of users are integrated to improve the overall performance of the recommendation system. The experimental results show that the system recommendation reaches 22.85 s based on deep learning algorithm, while the multimodal model only requires 0.124 s. Compared with deep learning algorithm, it can save 99.47% of time and greatly improve the speed of product recommendation. And the recommendation accuracy of multimodal models is always higher than that of deep learning algorithms. When the number of iterations is 100, the accuracy of multimodal models is 89.5%, an increase of 13.86% compared to deep learning algorithms. |
| keywords:multimodal model recommendation efficiency recommendation system memory accuracy rate |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|