地理社交网络中群体推荐算法的综述

The survey of group recommendation on Geo-Social Networks

随着移动互联网和智能手机的广泛应用,地理社交网络已经成为人们日常生活不可或缺的社交信息平台。由于人是一种具有典型社会属性的群体,群体活动是人在现实世界中的日常生活、工作的状态写照。因此,在这种新型的社交信息平台中,关于群体以及群体活动相关的推荐服务需求越来越成为当前推荐系统中的重要任务。虽然,目前已有大量研究来解决面向群体活动的群体推荐问题,但在群体推荐的研究中仍然存在一些问题:(1)群体活动决策过程中群体成员对决策的贡献程度不能直接从数据中获得;(2)现有的关于群体推荐的算法大都强调被推荐的群体是当前存在的,然而,事实上参加活动的群体大都是应活动的发起而组建的。因此,本文对地理社交网络中群体推荐算法的相关研究进行了系统地分析和整理;主要从当前的已有研究中采用的方法和研究相关的数据集两方面进行总结;最后,提出了该领域的几个潜在研究方向。

With the wide application of mobile Internet and smart phones,geographic social network has become an indispensable social information platform in people’s daily life.Since people are groups with typical social attributes,group activities are the portrayal of people’s daily life and work in the real world.Therefore,in this new type of social information platform,the demand for recommendation services related to groups and group activities has increasingly become an important task in current recommendation systems.Although a lot of research has been done on the group recommendation problem of group activities,there are still some problems in the research of group recommendation:(1)the contribution of group members to decision-making in the process of group activity decision-making cannot be directly obtained from the data;(2)the most existing groups recommendation algorithms emphasize that recommendation groups currently exist.However,most of the groups that participate in the campaign are realitiely formed in response to the campaign’s initiation.Therefore,this paper mainly systematically analyzes and organizes related researches on group recommendation algorithms in heterogeneous geographic social networks from the methods and datasets used in existing research.Finally,several potential research directions in this field are proposed.