挖掘用户兴趣的序列化推荐算法研究
首发时间:2023-02-14
摘要:本文通过对推荐算法现状进行研究发掘出用户兴趣对推荐结果的影响。之后通过推荐算法的推荐结果缓解用户的信息过载问题,用户能够更加快速准确地得到自己感兴趣的物品,同时提高物品提供平台的收益。本文方法的重点内容是在每个用户对应的序列化用户-物品交互数据集中,利用这些序列化的行为数据,进行数据的处理和构建,从中利用能够处理图形结构和序列结构的神经网络模型,并利用物品本身的类别特征,提取得到当前用户本身的兴趣,并通过得到的用户的兴趣对该用户之后的交互的物品进行预测,通过预测的准确性,得到用户兴趣提取的准确程度。最后通过实验,验证模型效果好过经典的推荐算法基线模型,并验证了本文模型的效果。
关键词: 管理科学与工程 序列化推荐 图神经网络 注意力机制 用户兴趣
For information in English, please click here
Research on sequential recommendation algorithm for mining user interest
Abstract:In this paper we studies the current situation of recommendation algorithm to explore the influence of user interest on recommendation results. Later, the recommendation results of the recommendation algorithm are used to alleviate the problem of users\' information overload, so that users can get the items they are interested in more quickly and accurately, and improve the revenue of the goods provision platform. The key content of the method of this paper is in the corresponding serialized user-item interaction data set, using the serialized behavior data to process and build the data, using the neural network model that can handle the graphic structure and sequence structure, and extract the interest of the current user itself, and predict the interaction of the user through the interest of the user, and the accuracy of the user. Finally, the effect of the model is better than the classic recommendation algorithm baseline model, and the effect of the model is verified.
Keywords: Management science and engineering Sequential recommendation Graph neural network Attention mechanism User interest
基金:
引用
No.****
同行评议
勘误表
挖掘用户兴趣的序列化推荐算法研究
评论
全部评论