面向多源异构数据的个性化搜索和推荐算法综述

Review on personalized search and recommendation algorithms for multi-source heterogeneous data

中文关键词  个性化搜索  多源异构数据  用户兴趣模型  深度学习

英文关键词  personalized search  multi-source heterogeneous data  user interest model  deep learning

中文摘要

      高效精准的个性化搜索、推荐等服务可为人们生产生活带来极大便利, 而随着互联网技术的迅猛发展, 面向多源异构数据的个性化搜索和推荐任务逐渐变得日趋复杂, 也是当前大数据分析及个性化服务领域的研究热点和难点. 个性化搜索和推荐算法广泛收集多源异构数据, 获取用户偏好信息, 利用各类机器学习、深度学习等技术,构建用户兴趣偏好模型, 预测用户偏好, 推荐满足用户个性化需求和偏好的项目或内容, 提升用户的使用体验和网站平台的商业利益. 本文介绍面向多源异构数据的个性化搜索问题的数学描述, 综述面向多源异构数据的个性化搜索和推荐算法的相关研究工作, 包括: 传统个性化搜索和推荐算法、融合多源异构数据的个性化搜索和推荐算法以及动态个性化搜索和推荐算法等相关研究现状, 整理了算法常用数据集、性能评价指标及评估体系, 进一步阐明了目前面向多源异构数据的个性化搜索和推荐方法的实际应用场景及今后研究的发展方向, 并讨论了存在的不足及所面临的严峻挑战, 期望为相关领域的研究人员提供有益帮助.

英文摘要

      Efficient personalized search service can bring great convenience in the production and life. With the rapid development of Internet technology, personalized search and recommendation task tends to become increasingly complex and is a hot research topic in the field of big data analysis. Personalized search and recommendation algorithms extensively collect user-generated content and obtain users’ preference information. By using various machine learning, deep learning and other technologies, these algorithms build user interest preference models, predict users’ behaviors, and recommend personalized items. It will improve users’ experiences and commercial benefits. This paper introduces the description of the personalized search problem, and reviews the research work on the personalized search and recommendation algorithms for multi-source heterogeneous data. It includes traditional personalized search algorithms, personalized search algorithms with multi-source heterogeneous data and dynamic personalized search algorithms. It sortes out common data sets and evaluation indicators, and clarifies the practical application scenarios and development directions of the personalized search methods for multi-source heterogeneous data. It also discusses the deficiencies and challenges, which is expected to be helpful to researchers in related fields.