面向SaaS的站内搜索模型的设计与实现

发布时间：2018-06-25 20:10

本文选题：SaaS + 站内搜索　；参考：《北京邮电大学》2017年硕士论文

【摘要】：站内搜索引擎作为网站平台内部的搜索引擎,通过分析用户输入并检索返回相关站内数据,为平台的用户提供了快速、便捷的数据访问入口。而基于SaaS的网站往往具有多租户、功能多样性、数据多样性和不同的用户权限等级等特征。通过设计和实现面向SaaS的站内搜索模型,不仅可以缩短SaaS服务提供商的开发周期,还能根据需要自主定制索引规则,具备灵活性和可拓展性。作者所在实验室也在开发和维护基于SaaS的CRM系统。因此作者从实际需求角度出发,研究并分析了当前站内搜索引擎的发展现状和SaaS模式的特点,规划了本课题的研究内容。在本课题中,作者对传统站内搜索引擎实现方案进行了比较详细的归纳总结和技术调研,提出了改进的方向。通过分析目前已有的站内搜索引擎的具体实现,从架构层次抽象地给出了一个完整站内搜索引擎应具有的模块以及模块之间的关系。在该模型中,设计了一种索引结构将SaaS平台的结构化和非结构化数据进行了统一。设计并实现了可配置相关索引规则的接口,可供配置的检索结果接口。除此之外,也将模型中其它接口进行了具体实现并将面向SaaS的站内搜索模型应用到了现有实验室平台中,通过实际应用,验证了架构和技术方面的可行性。在创新性方面,作者分析了以往的分布式索引方法,并在分布式B+树方法的基础上进行改进,提出了一种索引线性分块和B+树相结合的方法。并且在仿真实验中,验证了该方法在检索效率和插入效率方面的性能优势。
[Abstract]:As the internal search engine of the website platform, the in-site search engine provides a quick and convenient data access portal for the users of the platform by analyzing the user input and retrieving the relevant data. SaaS-based websites often have the characteristics of multi-tenant, function diversity, data diversity and different levels of user authority. By designing and implementing the in-station search model for SaaS, not only the development cycle of SaaS service provider can be shortened, but also the index rules can be customized according to the need, and it is flexible and scalable. The author's lab is also developing and maintaining SaaS-based CRM systems. Therefore, from the point of view of actual requirements, the author studies and analyzes the current status of search engine development and the characteristics of SaaS model, and plans the research content of this topic. In this paper, the author summarizes and summarizes the implementation scheme of traditional search engine in detail, and puts forward the direction of improvement. Based on the analysis of the implementation of the existing in-site search engine, this paper abstractly presents the module and the relationship between the modules of the whole search engine in the site from the architectural level. In this model, an index structure is designed to unify the structured and unstructured data of SaaS platform. The interface of configurable index rules is designed and implemented, and the retrieval result interface can be configured. In addition, the other interfaces in the model are implemented and the Saa-oriented in-station search model is applied to the existing laboratory platform. The feasibility of architecture and technology is verified by practical application. In the aspect of innovation, the author analyzes the previous distributed index methods, improves on the distributed B-tree method, and proposes a method of combining index linear partitioning with B-tree. In the simulation experiment, the performance advantage of the method in retrieval efficiency and insertion efficiency is verified.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.092;TP391.3

【参考文献】