摘要
为了在影视、书籍等单一类型推荐中准确地表达用户的真实偏好,充分地捕获到推荐数据中的有效特征,研究并提出一种融合门控注意力机制与双线性特征交互的推荐模型。使用融入门控机制的注意力单元来对用户的局部显性偏好建模,使用双线性特征交互层来对用户的长期泛性偏好进行挖掘,以提升深度推荐模型的学习能力。在Amazon(Books)和MovieLens-1M两个公开数据集中进行实验,实验结果表明所提模型相比于其他推荐模型,推荐效果有一定程度的提升。
In order to accurately express the users'true preferences in a single type of recommendation such as movies and books,and fully capture the effective features in the recommendation data,a recommendation model that integrates the gating attention mechanism and bilinear feature interaction is proposed.This model used the attention unit integrated into the gated mechanism to model the user's local explicit preferences and used bilinear feature interaction layer to mine the long-term general preferences of users to improve the learning ability of the deep recommendation model.Experiments were conducted on two public data sets,Amazon(Books)and MovieLens-1M.The experimental results show that the proposed model has a certain degree of improvement in recommendation effect compared with other recommendation models.
作者
何昌隆
文斌
He Changlong;Wen Bin(College of Communication Engineering,Chengdu University of Information Technology,Chengdu 610225,Sichuan,China)
出处
《计算机应用与软件》
北大核心
2024年第4期291-296,共6页
Computer Applications and Software
关键词
推荐系统
深度学习
注意力机制
双线性函数
多层感知机
Recommendation system
Deep learning
Attention mechanism
Bilinear function
Multilayer perceptron
