基于用户隐性反馈行为的下一个购物篮推荐

发布时间：2018-06-13 18:57

本文选题：下一个购物篮推荐 + 隐性反馈　；参考：《中文信息学报》2017年05期

【摘要】：下一个购物篮推荐是当前电子商务领域中极其重要的一项任务,传统的下一个购物篮推荐方法主要分为时序推荐模型和总体推荐模型。这些方法对点击、收藏、加入购物车等用户的隐性反馈行为利用得不够,并且没有考虑用户行为偏好的时间敏感性。该文提出了一种基于用户隐性反馈行为的下一个购物篮推荐方法,将用户行为按照一定的时间窗口进行划分,对于每个窗口从多个维度抽取用户对商品的时序偏好特征,运用深度学习领域的卷积神经网络模型进行分类器训练。在真实数据集中的实验结果表明,与传统的线性模型和树模型等分类器相比,该文提出的卷积神经网络框架具有较强的特征萃取能力和泛化能力,提高了推荐系统的用户满意度。
[Abstract]:The next shopping basket recommendation is an extremely important task in the field of electronic commerce at present. The traditional next shopping basket recommendation method is mainly divided into sequential recommendation model and general recommendation model. These methods do not take advantage of the implicit feedback behavior of users such as clicking, collecting and adding shopping cart, and do not take into account the time sensitivity of user's behavior preference. In this paper, the next shopping basket recommendation method based on the implicit feedback behavior of users is proposed. The user behavior is divided according to a certain time window. For each window, the temporal preference characteristics of the items are extracted from multiple dimensions. The classifier is trained by the convolution neural network model in the field of deep learning. Experimental results in real data sets show that the proposed convolution neural network framework has better feature extraction ability and generalization ability than traditional classifiers such as linear model and tree model. The user satisfaction of the recommendation system is improved.
【作者单位】：大连理工大学计算机科学与技术学院;
【基金】：国家自然科学基金(61572102,61632011,61562080) 国家重点研发计划(2016YFB1001103)
【分类号】：TP391.3

【相似文献】