Scalable Enterprise Systems Phase II:Exploratory Research on Flexible Production Networks: SEEK - Scalable Extraction of Enterprise Knowledge

摘要：CMS0122193 -“柔性生产网络第二阶段：SEEK -企业知识的可扩展提取”这项拨款为支持新理论和新技术的发展提供资金，以实现各种遗留系统与企业决策支持工具的无缝、快速互连，从而在这些遗留系统之间提取数据和知识。这种能力对于支持扩展企业协作的计算系统的快速配置是至关重要的。应用域是灵活的生产网络。灵活的生产网络（fpn）由在项目导向的定制生产环境中运营的公司组成。大多数建筑业（约占国民生产总值的8%）和不断增长的制造业都是以这种形式运作的。fpn的特点是根据项目需要快速组装和拆卸业务网络，特别适合于所提议的研究。我们称我们的技术为SEEK——企业知识的可扩展提取。知识提取是困难的，因为驻留在公司中的数据不能直接映射到决策支持系统所需的输入。现有的方法不支持SEEK的基本功能：(1)通过很少的编程设置快速连接到遗留数据和应用程序；(2)将源知识与源不可用的决策支持工具联系起来。我们提供这些功能的新方法是使用关于存储在应用程序中的信息类型的编码领域知识，以支持与这些应用程序的半自动连接。我们的方法是独立于应用程序的，支持从结构化和非结构化源提取遗留数据。这种方法本质上比构建特定于应用程序的工具更具可伸缩性和健壮性。我们研究的成功结论将显著提高速度并减少实例化具有异构源的扩展企业计算平台所需的人力输入量。

ABSTRACTProposal: CMS-0122193PI: Joachim Hammer Institution: University of FloridaDate: July 21, 2001ABSTRACT CMS0122193 - "Flexible Production Networks Phase II: SEEK - Scalable Extraction of Enterprise Knowledge"This grant provides funding to support development of new theory and technologies to allow the seamless, rapid interconnection of diverse legacy systems with enterprise decision support tools to allow extraction of data and knowledge across those legacy systems. Such a capability is critically important to allow rapid configuration of computing systems that support extended enterprise collaboration. The application domain is the flexible production network. Flexible production networks (FPNs) consist of firms that operate in a project oriented, customized production environment. Most construction (totaling approximately 8% of GNP) and growing parts of manufacturing industry operate in this form. Characterized by rapid assembly and disassembly of business networks based on project needs, FPNs are particularly well-suited users for the proposed research.We call our technology SEEK - Scalable Extraction of Enterprise Knowledge. Knowledge extraction is difficult because data resident in the firm does not map directly to the input needed by a decision support system. Existing approaches do not support SEEK's fundamental capabilities: (1) rapid connection to legacy data and applications with little programmatic setup; and (2) linking source knowledge with decision support tools not natively available to the source. Our novel approach to provide these capabilities is to use encoded domain knowledge about the types of information that would be stored in applications to enable semiautomatic connection to those applications. Our approach is application independent, supporting extraction of legacy data from both structured and unstructured sources. This approach is inherently more scalable and robust than building application specific tools. Successful conclusion of our research will significantly increase the speed and reduce the amount of human input necessary to instantiate an extended enterprise computing platform with heterogeneous sources.