实时数据库中的数据传递及其转换方法的研究与实现

【作者】 汤晓兵；

【导师】 贾智平；

【作者基本信息】 山东大学 ， 计算机应用技术， 2005， 硕士

【摘要】 传统的关系数据库系统旨在处理永久、稳定的数据,强调数据的完整性与一致性,较少考虑与数据处理相关联的时间限制。实时数据库与其相比有很大的不同,它是数据库和实时系统在概念、技术、方法和机制上的结合,是一种数据和事务都有显式定时限制的数据库。实时数据库系统能够提供高速、及时的实时数据服务,能够有效地集成异构控制系统,提供分布式的数据服务。它和关系数据库一起构成了数据支撑平台,对信息集成起着极其重要的作用。 本文系统地分析了实时数据库系统的体系结构、时间特性、事务处理、调度策略和并发访问控制等特征,提出了系统设计方案和实现方法。 利用实时数据库系统中的数据传输及其转换技术,不仅可以生成数据仓库,还可以用来检索不同结构的数据源中的数据、对数据进行复杂转换,并将数据存储到另一个数据源中,以便对数据进行析取、分析汇总,为数据仓库的海量数据存储和实时、快速的检索提供了强大的支持。本文结合作者参与研制的《建设厅执业资格注册中心综合业务系统》项目,从数据库的实时访问的角度出发,探讨了实时数据库系统中的数据传递及其转换的设计方法和实现技术,以及实时机制的实现方法,从而在时间约束条件下保证共享数据的一致性和完整性。 实时数据库系统中最为典型的问题是利用数据库技术的特点解决实时系统中的数据管理问题,为数据库系统提供时间调度和资源分配的算法,以及实时数据处理的各种方法。采用聚类分析方法,能不间断地对大量数据进行采集、选择、过滤、存储并及时响应各种不同的数据服务请求,得到从关系型数据库整合、转换到实时数据库的高效率的途径,为实时数据库系统提供了强大的数据处理能力。 本文根据项目中实时数据库系统的应用背景,针对现有用户访问事务聚类算法复杂度较高的问题,提出了一种新的用户事务聚类算法。首先对用户事务中访问过的页面浏览次数和时间进行综合统计,并定义了函数f将其映射为页面访问时间属性值;其次通过抽取事务的特征事务元素集,建立新的聚类对象集事务,实现对原有事务的降维。最后给出计算事务相似度的方法,并得到一个模糊相似度矩阵以实现事务聚类。该算法具有较高的准确性,计算复杂度低,扩展性好,可以广泛地应用于电子商务等领域。 本文项目中利用基于XML的Web Service技术,显著地提高了基于Web的应用程序间的互操作性,为实时信息采集、更新和发布等提供了更新的手段,从而更好地实现数据集成。更多还原

【Abstract】 The main function of a traditional relational database is to manage permanent and steady data. It focuses on the integrality and consistency and scarcely considers the time restriction of data management. Real time database differs greatly from the traditional database. It combines real time system with database on the concept, technology, method and mechanism. It is a kind of database in which data and transaction are time limited. The correctness of system depends on not only the logical result of the transaction but also the time when the logical results are produced. Real time database can provide the rapid and timely data services, integrate the heterogeneous control system effectively, and offer the distributed data services. Together with the relational database, the real time database constitutes the data platform and plays an important role in the integration of information.After systematically analyzing the characteristics of the architecture, the time characteristic, transaction management, scheduler politics and concurrent access control, etc., the paper puts forward the design scheme and implement methods of the system.The real time database technology of data transfer and conversion is not only used to produce the data warehouse but also search the data in different structure data sources, transfer the data and restore the data in another data source in order to extract, analyze and gather the data. In that way, it supports data storage and rapid real time search strongly. According to the collaboration project(The integrative business system of Practice Qualification Registration Center of BUREAU OF CONSTRUCTION), the author discusses the design method of data transferability and transformation and applies the technology of data transferability and transformation and realization method of the real time mechanism in view of real time access. As a result, the consistence and the integralityof the shared data are ensured by the time restriction.The most typical problem of the real time database is solving the data management in real time database, providing the algorithm of time schedule and resource allocation and giving the various methods of managing real time data. The clustering analysis method can collect, select, filtrate, store the large account of data continually and response to different data service in time .It has an effective way to integrate the relational database with real time database and transform the relational database to real time database. It provides a tremendous capability of data processing for the real time database system.According to applying background of real time database system and the problem of higher complexity in the clustering algorithm of user accessing schedule, the paper puts forward a new clustering algorithm. Firstly, we synthesize the statistic of the times and time of users’ visiting page which mapped to visiting time attribute value. Then, the dimension is descended by extracting the feature elements from a original user session. Finally, we give a similarity method of session and get a fuzzy similarity matrix which can cluster session. This algorithm is proved to have better accuracy, fewer CPU time and better scalability than others by the experiments. It can be widely used in the fields of Electronic Commerce.This project has applied the technology of Web Service based on XML. This technology improves the interactivity of the application based on Web and provides a new method of the collection, update and release of real time information, which implements the integration of information greatly.更多还原