基于Lucene的分布式全文检索在构件库系统中的实现-软件工程专业论文.docxVIP

摘要 随着软件复用与基于构件的软件开发实践的不断深入和发展，以及分布式软 件构件库规模的不断扩大，在分布式构件库中检索符合需求的构件变得非常困难。 因此针对分布式构件库系统建立一个快速有效的分布式检索系统是非常必要的。 本文从分布式构件库系统的特点出发，深入剖析了全文检索引擎 Lucene 的组 织结构、索引文件结构以及工作原理。并对多种中文分词工具做出了分析，选用 了适合构件库系统的中文分词工具。由于 Lucene 仅能处理纯文本格式数据。因此， 本文在 Lucene 中增加多格式文档处理模块，以实现对多种文件格式的支持。并针 对分布式构件库的特点，制定了异步通信框架，以实现分布式的全文检索。本文 在设计上以面向对象的思想为指导，实现了全文检索与分布式构件库系统的松耦 合，便于全文系统向其他系统的移植，并采用多种设计模式，易于开发人员对系 统的扩展和维护。 最后本文通过测试证明，本文针对分布式构件库特点，构建的分布式全文检索 系统在查全率和查准率以及响应时间等指标上都达到了预期的目标，也完全可以 满足用户的需求。 关键词：构件库 Lucene 全文检索 Abstract With software reuse and component-based software development unceasingly deepens and develops, and the scale of distributed software component library is increasingly expanding, which makes it difficult to retrieve the needed components in distributed component library. Therefore, it is quite necessary to establish a fast and effective distributed retrieval system for distributed component library. This paper, starting from the features of distributed component library system, makes a deep analysis on the organization structure, index file structure and working principle of full-text retrieval engine Lucene. This paper also conducts an analysis on a variety of Chinese word segmentation tools and selects those tools which are suitable for component library. However, Lecene could only process the data in plain text. Thus, the processing modules for multi-format documents are added in Lucene in this paper, so as to support multiple document formats. In addition, asynchronous communication frame is established based on the features of distributed component library to realize distributed full-text retrieval. By taking object-oriented thought as guidance in design, this paper realizes the loose coupling of full-text retrieval and component library system, which allows the full-text system to be transplanted to other systems in a convenient way, and various design modes adopted in this paper make it easier for developers to expand and maintain the systems. It is finally proved by test th