SoD: Incremental Design of Complex Information Systems

如今，用于设计小型个人数据库的工具通常与用于设计企业级信息系统的工具一样复杂。这对人类有效处理这些工具中使用的复杂结构的能力提出了挑战。此外，随着世界变得越来越网络化，我们发现访问和集成来自多个自主创建的数据存储的信息是可能的，也是可取的。即使这些数据存储中的每一个都很简单，整个集合也可能非常复杂。最后，信息存储会随着时间的推移而持续存在，而我们的需求也会在这段时间内发生变化。因此，商店需要与时俱进，我们对它们的查询也是如此。信息系统的设计必须适应这种变化。该项目开发了基于XML的信息系统的增量设计技术。开发了用于模式的分层分解的技术，并通过此来解析被视为相关信息上下文的“视图”的多个不完整的模式规范。还开发了用于将这些映射到XML模式的自动化技术，该技术既适合于高效的查询处理，又被标准化以满足已知的函数依赖并避免更新异常。这个项目为复杂的异类信息系统的设计开发了一个正式的概念基础。它还开发了增量技术来处理随时间推移的演变变化，以及数据存储的合并(和丢失)。如果成功，这个项目将大大减少设计和维护信息存储的人力，特别是对于可能不熟悉基本技术细节的信息系统的设计者和用户，如领域专家，他们对从许多不同的信息源收集信息感兴趣，而且他们的信息需求变化很快。通过与一群生物学家密切合作，这种影响的可能性将最大化，这些生物学家将作为该技术的当地用户并提供反馈。项目网站http://www.eecs.umich.edu/~jag/design/将用于传播结果，包括在该项目中创建的软件工具，我们将免费提供给学术使用。

Often today, the tools used to design a small personal database are as complex as the tools used to design enterprise-size information systems. This challenges the human ability to deal effectively with the complex structures employed in these tools. In addition, as the world gets increasingly networked, we find it possible and desirable to access, and integrate information from multiple autonomously created data stores. Even if each of these data stores individually is simple, the entire collection may be very complex. Finally, information stores persist over time, and our needs change over this period. In consequence, the stores need to evolve with time, as do our queries of them. Information systems must be designed to adapt to this change. This project develops techniques for the incremental design of XML-based information systems. Techniques are developed for the hierarchical decomposition of schemas and through this, resolution of multiple incomplete schema specifications considered as "views" of related information contexts. Automated techniques are also developed for mapping these into XML schema that are both suited for efficient query processing as well as being normalized to satisfy known functional dependencies and avoid update anomalies. This project develops a formal conceptual basis for the design of complex heterogeneous information systems. It also develops incremental techniques to deal with evolutionary change over time, as well as with the merger (and loss) of data stores. If successful, this project will greatly reduce the human effort involved in the design and maintenance of information stores, particularly for designers and users of information systems who may not be familiar with the underlying technical details, such as domain experts who are interested in bringing together information from many heterogeneous information sources and whose information needs change rapidly. The likelihood of such impact will be maximized by working closely with a group of biologists who will serve as local users of the technology and provide feedback. The project Web site http://www.eecs.umich.edu/~jag/design/ will be used to disseminate the results, including software tools created in this project that we will make available free of charge for academic use.