【作者】 周煜；

【导师】 胡雯蔷；

【作者基本信息】 华中科技大学 ， 软件工程， 2020， 硕士

【摘要】 随着多媒体技术及互联网技术的发展和普及,线上数字化娱乐已经迅速融入了人们的生活中,技术深深地改变了数字娱乐产品的生产方式与人们的消费娱乐模式。于此同时,数字化娱乐产品的生产方式与新的消费模式的形成,极大提高了数字化娱乐产品的生产效率,各种影音娱乐产品呈现爆发式增长的趋势。面对庞大的信息海洋,如何利用新技术智能地为人们选择适合自己的包括娱乐产品在内的各种产品成为现实的需求。在这种时代背景下,推荐系统成为为人们提供信息匹配服务的热门工具。针对这个问题,论文实现了一个追求高可用性和高稳定性的音乐推荐系统。首先,论文对目前广泛应用的推荐算法的优缺点进行了深入的理解和讨论,通过分析协同过滤算法中存在的数据稀疏性问题、冷启动问题、可伸缩性问题和计算复杂度问题,论文实现了基于因子分解机等推荐算法的音乐推荐系统。该算法能够在稀疏性环境下进行学习,并且有效的降低了计算复杂度,解决了数据稀疏性问题和伸缩性问题。针对冷启动问题,系统融合了基于内容的推荐和基于热度的推荐算法等丰富了系统考虑的推荐角度。同时,论文应用了微服务的设计思想,对系统功能进行拆分,给出了各个模块的设计,使用Springboot框架、Spring Cloud Alibaba解决方案的部分组件、Vue JS框架以及FM算法在内的多种推荐算法等技术,实现了音乐推荐系统。经过系统化测试,该系统运行正常,能够精准地为客户进行个性化推荐,系统的各个部分互相独立,并能通过RPC调用完成设计功能,满足了用户选择音乐的需求。更多还原

【Abstract】 With the development and popularization of multimedia technology and Internet technology,online digital entertainment has quickly integrated into people’s lives,and technology has profoundly changed the production methods of digital entertainment products and people’s consumption and entertainment models.At the same time,the production methods of digital entertainment products and the formation of new consumption models have greatly improved the production efficiency of digital entertainment products,and various audio-visual entertainment products have shown an explosive growth trend.Faced with a huge ocean of information,how to use new technologies to intelligently select suitable products for people,including entertainment products,has become a real demand.In this era,the recommendation system has become a popular tool for people to provide information matching services.In response to this problem,the paper implements a music recommendation system that pursues high availability and high stability.First of all,the paper has an in-depth understanding and discussion of the advantages and disadvantages of the widely used recommendation algorithm.By analyzing the data sparsity problem,cold start problem,scalability problem and computational complexity problem in the collaborative filtering algorithm,the paper realize A music recommendation system based on factorization machine and other recommendation algorithms is introduced.The algorithm can learn in a sparse environment,and effectively reduces the computational complexity,and solves the problem of data sparsity and scalability.For the cold start problem,the system incorporates content-based recommendation and heat-based recommendation algorithm to enrich the system’s recommendation perspective.At the same time,the paper applies the design concept of microservices,splits the system functions,and gives the design of each module,using the Springboot framework,some components of the Spring Cloud Alibaba solution,the Vue JS framework,and FM algorithm,including a variety of recommendations Algorithms and other technologies have realized a music recommendation system.After systematic testing,the system operates normally and can accurately recommendpersonalized recommendations for customers.Each part of the system is independent of each other,and can complete the design function through RPC calls,which meets the user’s needs for selecting music.更多还原