基于偏好特征提取的农产品电商信息个性化推送方法
Personalized Push Method for Agricultural Product E-Commerce Information Based on Preference Feature Extraction

由于现行方法在农产品电商信息个性化推送中存在用户偏好理解不够深入、推送算法不精准等，导致应用效果不佳。具体表现为推送内容与用户实际需求存在偏差，用户满意度不高，同时，由于推送算法的性能限制，导致推送准确性较差，无法达到预期的推送效果。针对农产品电商信息个性化推送中现行方法存在的用户偏好理解不足、推送算法精准度低等问题，本文提出了一种基于偏好特征提取的个性化推送方法。该方法通过深入分析电商平台上的用户反馈信息，区分正负维度并提取正样本，进一步从正样本中提取用户偏好特征。基于这些偏好特征，我们对农产品进行推送评分，并生成个性化推送名单，在电商平台上实施推送。实验结果显示，该方法在MRR和HR两个评价指标上均不低于0.8，表现出较高的推送精度，有效解决了现有推送方法中的不足，实现了农产品电商信息的个性化精准推送。
Due to the lack of in-depth understanding of user preferences and inaccurate push algorithms in the current method of personalized information push for agricultural product e-commerce, the application effect is not satisfactory. Specifically, there is a deviation between the pushed content and the actual needs of users, resulting in low user satisfaction. At the same time, due to the performance limitations of the push algorithm, the accuracy of the push is poor, and the expected push effect cannot be achieved. This paper proposes a personalized push method based on preference feature extraction to address the problems of insufficient understanding of user preferences and low accuracy of push algorithms in the current methods of personalized push of agricultural product e-commerce information. This method analyzes user feedback information on e-commerce platforms in depth, distinguishes positive and negative dimensions, and extracts positive samples to further extract user preference features from the positive samples. Based on these preference characteristics, we conduct push ratings on agricultural products and generate personalized push lists for implementation on e-commerce platforms. The experimental results show that this method achieves a high push accuracy of not less than 0.8 on both MRR and HR evaluation indicators, effectively solving the shortcomings of existing push methods and realizing personalized and accurate push of agricultural product e-commerce information.