基于深度学习的会话推荐方法综述  

Review of Deep Learning-based Conversation Recommendation Methods

作　　者：袁凤源 梅红岩[1] 温民伟 白杨 吴帅甫 YUAN Feng-yuan;MEI Hong-yan;WEN Min-wei;BAI Yang;WU Shuai-fu(School of Electronics&Information Engineering,Liaoning University of Technology,Jinzhou 121001,China)

机构地区：[1]辽宁工业大学电子与信息工程学院,辽宁锦州121001

出　　处：《辽宁工业大学学报（自然科学版）》2024年第1期6-10,17,共6页Journal of Liaoning University of Technology(Natural Science Edition)

摘　　要：随着信息技术和智能应用的迅速发展,现今社会面临着大量数据和信息过载的挑战。为解决这一难题,基于会话的推荐方法应运而生。基于深度学习的会话推荐方法利用其强大的表示学习能力,更准确地预测用户的短期兴趣,并提供个性化的推荐服务。故综述了深度学习会话推荐方法的研究进展,包括基于卷积神经网络、图神经网络、注意力机制、多层感知器、混合模型等方法。总结了研究难点和未来研究方向。随着深度学习技术的不断进步,会话推荐方法将在实际应用中发挥越来越重要的作用。As information technology and intelligent applications continues to advance quickly,today’s society is facing challenges such as data and information overload.To solve this problem,conversation-based recommendation methods have emerged.Based on deep learning,these methods use their powerful representation learning ability to more accurately predict users’short-term interests and provide personalized recommendation services.This article reviews the research progress of deep learning-based conversation recommendation methods,including methods based on convolutional neural networks,graph neural networks,attention mechanisms,multi-layer perceptrons,and hybrid models.The article summarizes the research challenges and future directions.As deep learning technology continues to make progress,conversation-based recommendation methods are expected to become increasingly important in practical applications.

关 键 词：会话 推荐系统 深度学习 机器学习 神经网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]