Research on Distributed and Evolutional Database Systems

本研究的目的是开发具有动态重组能力的新数据库系统技术，以整合网络上各种类型的数据，并允许每种数据结构的演变。本文主要研究成果如下：(1)多媒体数据的组织。考虑到视频/图像数据的主要特征（数据连续性），我们开发了一种代数视频检索模型、基于图论方法和人工智能方法（故事语法）的视频创作系统、基于内容封装的版权保护和收费管理系统、一种新的具有连续细节层次的多媒体数据访问控制方法。一种基于语音/图像/字符识别技术的运动视频场景自动分类方法。(2)互联网数据库技术：基于自组织映射的Web数据自动分类方法，基于动态决策树的信息过滤方法，基于路径存在约束的超链接一致性控制方法，基于数据挖掘中的关联规则的Web数据检索引擎。(3)移动和便携数据库为了解决移动通信中的几个问题，我们利用主动数据库机制开发了一个移动数据库系统。此外，我们还开发了一种利用位置信息和数据副本的高效查询处理方法。(4)分布式代理数据库为了在开放网络中集成异构信息资源（如半结构化和多媒体数据），我们采用代理技术（KQML）、分布式对象管理技术（CORBA）、ECA规则技术以及演绎和面向对象技术开发了中介系统。

The aim of this research is to develop new database system technologies with dynamic reorganization capabilities, which integrate a variety of types of data on networks, and allow the evolution of each data structure. The following are the major results of the research project :(1) Organization of multimedia dataConsidering the major characteristic of video/image data (data continuity), we developed an algebraic video retrieval model, video authoring systems by graph-theoretical approach and AI approach (story grammer), a copyright protection and a charge management system by content capsulation, a new data access control method for multimedia data with continuous levels of details, an extracting and indexing method by speech/image/character recognition technologies, an automatic classification method for sport video scenes.(2) Internet database technologiesWe developed an automatic classification method of Web data by the Self-organizing map, an information filtering method by the dynamic decision tree, a consistency control method for hyperlinks by the path existence constrainta, and a retrieval engine for Web data by association rules of data mining technologies.(3) Mobile and portable databasesWe developed a mobile database system by the active database mechanism in order to cope with several problems in mobile communication. Furthermore, we developed an efficient query processing method by using location information and data replica.(4) Distributed agent databasesIn order to integrate heterogeneous information resources (e. g. semistructured and multimedia data) in open networks, we developed mediation systems by the agent technology (KQML), the distributed object management technologies (CORBA), the ECA rule technology, and the deductive and object-oriented technology.