序列推荐算法的研究综述

Survey of Research on Sequential Recommendation

随着网络技术的飞速发展,信息量呈现出爆发式增长的态势,推荐算法也随着信息量的激增获得了长足的发展。序列推荐从用户行为出发,目的得到更为准确的动态推荐效果。序列推荐算法主要分为两类,一类为基于传统推荐算法的序列推荐,例如马尔科夫模型和协同过滤,另一类为基于深度神经网络的序列推荐算法。近些年,越来越多深度神经网络融入序列推荐算法,最近,由于图网络的信息覆盖更为全面,图网络也渐渐融入到序列推荐算法中。

With the rapid development of online technology, the information shows an explosive growth trend, and recommendation algorithms have also made great progress with the surge of information. Sequential recommendation starts from user behavior and aims to obtain more accurate dynamic recommendation effect. Sequential recommendation algorithms are mainly divided into two types, one is sequential recommendation based on traditional recommendation algorithms, such as Markov model and collaborative filtering, and the other is sequential recommendation algorithm based on deep neural network. In recent years, more and more deep neural networks have been integrated into sequential recommendation algorithms. Recently, due to the more comprehensive information coverage of graph networks, graph networks have gradually been integrated into sequence recommendation algorithms.