面向新领域的推荐系统综述

Survey of recommendation systems for new domains

互联网时代的快速演进带来了数据信息的海量增长,推荐系统旨在海量数据中提升用户获取信息的有效性。同时推荐系统促进领域的发展并带来了新的机遇,现阶段各行业在应用中涌现出大量新领域下为用户进行个性化推荐的需求。然而,推荐系统在新的领域进行实际场景应用,往往需要从零开始构架数据体系,这依赖专家对领域特征进行分析,总结获取新领域的数据关系,并且存在推荐系统应用领域限定性强、数据稀疏、冷启动等阻碍。本文旨在面向新领域推荐系统的构建这一主题进行综述,从新领域背景和挑战、新领域推荐方法、方法评估3个研究方向,介绍了相关工作的研究现状,给出了研究面向新领域技术应用的建议,并对新领域推荐的发展趋势进行展望,指出下一步需要开展的工作。

Nowadays,the rapid evolution of the Internet era has brought about the massive growth of data and information,and the recommendation system aims to improve the effectiveness of users'access to information in the massive data.At the same time,recommendation systems promote the development of the domain and bring new opportunities,and at this stage,there are a lot of demands for personalized recommendations for users in new domains emerging from various industries in the application.However,recommendation systems for practical scenario applications in new domains often require framing the data system from scratch,which relies on experts to analyze domain features and summarize to obtain data relationships in new domains,and there are hindrances such as strong limitation of recommendation system application domains,sparse data,and cold start.This paper aims to review the topic of building recommendation systems for new domains.Specifically,it introduces the current research status of related work in three research directions:background and challenges of new domains,research on new domain recommendation methods,and evaluation of methods.This paper gives suggestions for researching the application of technologies oriented to new domains and gives an outlook on the development trend of new domain recommendation and points out the next work to be carried out.