服务搜索引擎中个性化服务推荐功能的设计与实现

发布时间：2018-01-20 21:43

本文关键词： Web服务推荐系统协同过滤语义分析　出处：《北京邮电大学》2013年硕士论文　论文类型：学位论文

【摘要】：SOA因其良好的扩展性和灵活性得到了日渐广泛的使用。作为SOA的主要实现方式,Web服务也日益受到服务开发者们的重视。Web服务可以被灵活地集成到开发者自身的业务逻辑之中,极大地降低了业务开发的成本,提高了开发的效率。因此,开发人员急迫地期盼能够快速地发现其所需的Web服务。然而,目前基于UDDI的Web服务发现方式已经难以满足用户的需求,这严重地降低了SOA开发的效率。本文提出了建立在Web服务搜索引擎之上的服务推荐系统,利用协同过滤等推荐方法,从用户的数据中发掘智慧,帮助开发者们发现想要的Web服务,以提高SOA的开发效率。本文提出的Web服务推荐子系统是Web服务搜索引擎中的一个核心模块。Web服务推荐子系统针对四个典型的应用场景,采取了不同的推荐算法,向用户作个性化的服务推荐。笔者采用改进后的牛顿冷却定律算法进行热门服务推荐。同时,笔者改进了基于用户和基于物品的协同过滤算法,分别使用基于浏览历史的协同过滤算法和基于标签的语义相关度算法来减轻冷启动带来的问题。笔者还提出了基于标签分类的推荐算法来进行个性化的服务推荐。最后,笔者还设计了QoS过滤及排名算法,以此来保证个性化推荐服务的质量。本文首先研究了商用推荐系统及Web服务推荐系统研究的现状介绍了推荐系统中常用的技术和方法。在此基础之上,对Web服务推荐子系统的需求进行了详细的分析,将推荐目标定位为个性化的广度推荐。然后,本文详细地介绍了Web服务推荐子系统的设计以及实现。在此之后,针对系统的各个模块进行了详细的单元测试,同时,也对整个系统进行了系统测试,以验证设计的合理性。最后,本文对Web服务推荐子系统进行了工作总结,并且提出了尚需完善的地方。
[Abstract]:SOA is widely used because of its good expansibility and flexibility. It is the main way to implement SOA. Web services are increasingly valued by service developers. Web services can be flexibly integrated into the developer's own business logic, greatly reducing the cost of business development. As a result, developers urgently expect to be able to quickly discover the Web services they need. At present, the Web service discovery method based on UDDI has been difficult to meet the needs of users. This seriously reduces the efficiency of SOA development. This paper proposes a service recommendation system based on Web services search engine, using collaborative filtering and other recommendation methods to extract wisdom from users' data. Help developers find the desired Web services to improve the efficiency of SOA development. The Web services recommendation subsystem proposed in this paper is a core module in the Web services search engine. The web services recommendation subsystem adopts different recommendation algorithms for four typical application scenarios. The author uses the improved Newton's cooling law algorithm to recommend the popular service. At the same time, the author improves the collaborative filtering algorithm based on users and articles. The cooperative filtering algorithm based on browsing history and the semantic correlation algorithm based on label are used to alleviate the problem caused by cold start, and the recommendation algorithm based on label classification is put forward to carry out personalized service recommendation. ... finally. The author also designs QoS filtering and ranking algorithm to ensure the quality of personalized recommendation service. This paper first studies the current situation of commercial recommendation system and Web service recommendation system, and introduces the common technologies and methods in recommendation system. The requirement of Web service recommendation subsystem is analyzed in detail, and the recommendation target is defined as personalized breadth recommendation. Then. This paper introduces the design and implementation of Web service recommendation subsystem in detail. After that, the unit test is carried out for each module of the system, at the same time, the whole system is also tested. In order to verify the rationality of the design. Finally, this paper summarizes the work of Web services recommendation subsystem, and puts forward some points that need to be improved.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP391.3

【相似文献】