新闻推荐算法研究综述

Overview of research on news recommendation algorithms

随着互联网蓬勃发展,众多网络新闻不断产生,导致新闻信息过载,读者很难在海量新闻中找出自己感兴趣的内容,因此迫切需要有效的新闻推荐算法解决此问题。在介绍了新闻推荐算法的作用及其发展历程后,首先,对新闻推荐算法各模块组成进行介绍;其次,根据算法特性将其划分为传统的新闻推荐算法和基于深度学习的新闻推荐算法,对于基于深度学习的新闻推荐算法又将其划分为有无附加信息的新闻推荐算法,之后又将有附加信息的新闻推荐算法划分为基于时间、位置、社交以及会话的新闻推荐算法,并分别对其进行介绍;然后,归纳了新闻推荐算法的数据集以及各种评价指标。最后,对新闻推荐领域未来可能的发展以及研究方向进行了展望。

With the vigorous development of the Internet,many online news are constantly generated,which leads to news information overload.It is difficult for readers to find the content they are interested in in the massive news.Therefore,effective news recommendation algorithms are urgently needed to solve this problem.After introducing the role and development process of news recommendation algorithms,firstly,the composition of each module of news recommendation algorithms is introduced;secondly,according to the characteristics of the algorithm,it is divided into traditional news recommendation algorithms and news recommendation algorithms based on deep learning.For news recommendation algorithms based on deep learning,they are further divided into news recommendation algorithms with and without additional information.Afterwards,news recommendation algorithms with additional information are further divided into news recommendation algorithms based on time,location,social,and conversation,and introduced separately.Then,the dataset of news recommendation algorithms and various evaluation indicators were summarized.Finally,the possible future development and research directions in the field of news recommendation were discussed.