III-COR: Collaborative Research: The Morpheus Data Transformation Management System

这一研究项目的目标是通过促进独立构建的数据库的整合，促进跨企业边界的信息共享。因为这样的数据库从来没有相同的内容，所以有必要编写转换(或适配器)来将这种完全不同的数据转换成一种通用的形式。人们普遍认为，这种转换的构建是数据集成项目的主要成本。Morpheus项目的目的是通过在Web上爬行公开可用的转换并为高效的转换构建提供高级工具来在存储库中捕获大量这样的转换。此外，预计强大的浏览工具允许用户通过提供文档的关键字搜索、在转换的分类层次内搜索、按转换的出处搜索、以及按输入/输出特征搜索来在存储库中快速定位“感兴趣的”转换。预计Morpheus将大幅降低编写和维护数据集成转换的成本，这将缓解未来数据集成项目的难度。该项目将支持麻省理工学院和佛罗里达大学的研究生。此外，在这两个机构的数据库课程中，将使用转换构造作为学生练习。可从项目网站http://www.cise.ufl.edu/~jhammer/morpheus/获得更多信息，在那里将传播研究成果，并提供原型代码。

The goal of this research project is to facilitate the sharing of information across enterprise boundaries, by facilitating the integration of separately constructed data bases. Because such data bases never have identical content, it is necessary to write transforms (or adaptors) to convert such disparate data into a common form. The construction of such transforms is widely believed to be a major cost of data integration projects. The purpose of the Morpheus project is to capture a large number of such transforms in a repository by crawling the web for publicly available ones and providing high level tools for efficient transform construction. In addition, powerful browsing tools are anticipated that allow users to locate "interesting" transforms quickly in the repository by providing keyword search of documentation, search within a classification hierarchy of transforms, search by the provenance of transforms, as well as search by the input/output characteristics. Morpheus is expected to dramatically reduce the cost of writing and maintaining data integration transforms, which will ease the difficulty of future data integration projects. This project will support graduate students at both Massachusetts Institute of Technology and University of Florida. In addition, transform construction will be used as student exercises in data base classes at both institutions. Further information can be obtained from the project web site: http://www.cise.ufl.edu/~jhammer/morpheus/ where research results will be disseminated and prototype code will be available.