| 跨境电商下基于商品属性-情境的推荐算法 |
| Recommendation algorithm Based on commodity attributes-context under cross-border e-commerce |
| 摘要点击 838 全文点击 0 投稿时间：2021-03-04 修订日期：2023-01-20 |
| 查看/发表评论 下载PDF阅读器 |
| 中文关键词 跨境电商; 协同过滤; 推荐系统; 商品属性; 情境权重 |
| 英文关键词 Cross-border e-commerce; collaborative filtering; recommendation system; commodity attributes; context weight |
| 基金项目 国家自然科学基金项目（青年项目，面上项目，重点项目） |
| 投稿方向 |
| 作者 | 单位 | 邮编 | | 蔡学媛* | 武汉纺织大学 | 430200 | | 李建斌 | 华中科技大学 | | | 钱自顺 | 华中科技大学 | | | 戴宾 | 武汉大学 | |
|
| 中文摘要 |
| 在传统协同过滤推荐算法的基础上, 结合跨境电商行业特点, 提出了一种基于商品属性和情境权重的混
合推荐算法. 首先, 根据用户商品的历史购买金额和购买次数生成用户偏好评分, 并结合用户商品属性相似度和情
境化用户相似度得到目标用户的最近邻集, 最后将通过变异系数法得到的情境权重纳入评分预测当中, 进而生成推
荐结果. 实证分析表明, 本算法能有效提升商品推荐结果的预测准确度, 相较基于商品属性的协同过滤推荐算法, 本
算法可降低商品预测评分平均绝对误差平均达2.72%, 提高了跨境电商商品推荐效果. 本研究为推荐系统在跨境电
商领域的应用提供了新方法. |
| 英文摘要 |
| Based on traditional collaborative filtering recommendation algorithm and the characteristics of the cross-border e-commerce industry, this paper proposes a hybrid recommendation algorithm based on commodity attributes and context weights. First, generate user scores based on user’s historical purchase amount and frequency, and combine user product attribute similarity and context user similarity to obtain user’s nearest neighbor set, and finally the context weight is incorporated into the score prediction to get the recommendation results. The results show that this algorithm effectively improves the prediction accuracy of the product recommendation results. Compared to the collaborative filtering algorithm based on commodity attributes, the recommendation algorithm proposed can reduce the average absolute error of the product prediction score by an average of 2.72%, which improves the efficiency of cross-border e-commerce product recommendation. The research provides a new method for the application of recommendation systems in the field of cross-border e-commerce. |
| 关闭 |
|
|
|
|
|
