Manufacturing Enterprise Integration

Program Manager: Al Jones
Total FTE: 10

Program Goal:

By 2005, demonstrate the potential for reducing the cost of Business-to-Business (B2B) software integration through new types of semantics-based measurements, standards, and infrastructural technologies that enable the building and testing of self-integratable applications.

Customer Needs:

Starting in the mid 1980s, a new approach to integration appeared that is still in use today. That approach is to decompose integration into two parts: communication protocols and interface specifications. Communication protocols govern the physical exchange of bits and bytes between the computers on which the software applications execute. There are many national and international standards for these protocols. These standards can now be implemented in computer hardware and system software that is separate from the manufacturing applications. Interface specifications govern the syntax and semantics of every piece of information exchanged by those applications. For any particular specification, there may be none, one, or many standards. The availability of such standards has potential to reduce the number of required translators from O(N2) to O(N) — where O(.) means "on the order of", and N is the number of computers.

Largely, the benefits from the communication standards that govern Internet and network communications have been realized. Bits and bytes can be sent cheaply, quickly, securely, and accurately, from one computer to another anywhere in world. Similar benefits have not accrued from standardizing interface specifications for information exchanges. The development, testing, and implementation cycle for these specifications can take years. Furthermore, since numerous organizations and consortia develop them, multiple, even conflicting, standards arise for the same information. This leads to costly and time-consuming harmonization efforts, which produce longer and more costly cycles.

As manufacturing enterprises implement new business structures, they are making the Internet a critical part of their business strategy. The Internet makes it possible for manufacturers to actually link up with their partners, suppliers, and customers. However, the Internet is not enough. To turn possibility into reality requires automatic integration of a myriad of enterprise-level software applications.

Several market analysts, including Gartner and D.H Brown, project that the market for these applications will reach $100 B by the year 2005. They also estimate the integration costs to range from two to five times the software costs. A recent report from the Integrated Manufacturing Technology Roadmap (IMTR, http://imtr.ornl.gov/) briefly describes a new approach in which software applications negotiate interfaces automatically. The authors called this approach self-integration. A recent report from the RAND Corporation on integration and interopoerability (http://www.rand.org/publications/MR/MR1215/MR1215.chap6.pdf) contained a section titled "NIST's Evolving Role". The following excerpt is taken directly from that section.

"Are there ways of ensuring interoperability with lighter standards that do not have to specify as much, or, better yet, with translators and mediators that can dispense with many higher-level standards altogether? NIST itself can develop the parameters, corpora, tests, and testbeds that help measure the quality and fitness of ontologies and mediators."

A number of organizations, including Open Applications Group (OAG), RosettaNet, and Automotive Industry Action Group (AIAG) approached NIST about an interoperability testbed. The idea for this testbed facility came from a recent, one-time event (http://www.openapplications.org/) called the Vendor Challenge. This event attracted more than 20 vendors and three major manufacturers — Boeing, Lucent, and Lockheed Martin. Each manufacturer provided a supply-chain integration scenario that the vendors used to demonstrate interoperability of their software tools based on the OAG interoperability specifications. Unlike the Vendor Challenge, which was a one-time event, this testbed is expected to be persistent. As such, it will have an on-going testing and experimenting capability and will be open for use to a large collection of potential stakeholders.

Technical Approach:

Our technical research strategy has been to start with a simple supply chain scenario involving two software agents: a buyer agent, A1, and a seller agent, A2. We assume initially that A1 and A2 share a common ontology ONT-0. ONT-0 contains the semantics of some very basic terms related to buying and selling, and generic part names associated with the products to be bought and sold. In addition, each agent will have its own specialized ontology: ONT-1 will give the semantics of products to order for A1, and ONT-2 will give the semantics for the product catalog for A2. We assume that terms in ONT-1 and ONT-2 are defined on top of ONT-0 and thus can be reduced to terms in ONT-0. Initially, we will be using the DARPA agent markup language (DAML) to define these ontologies; in our later research, we will also use the process specification language (PSL).

Determining whether two software applications have the same semantic understanding of an information object is difficult. In a supply-chain environment, it is far more likely that semantically similar terms from two ontologies will agree on many, but not all, properties. Therefore, approximate or partial matching becomes a necessary part of the reasoning process. That process contains three major steps: measuring, negotiating, and mapping. In step one, measuring, we will develop a metric, a theory, and test methods, for comparing two information objects quantitatively. We will assess the applicability of the information uncertainty measures developed by Shannon, Stonier, and Zadeh. In addition, we will examine a numbre of quantitative and qualitative approaches to compute tehse measures. Step two, negotiating, will be done in conjunction with step one. We will use and expand the protocols being developed in ISO Technical Committee 184/Sub-committee 5/Working Group 1. In step 3, mapping, we must develop syntactic and semantic translators between the two objects. We plan to continue the earlier work on mappings conducted as part of PSL.

The research to achieve the ultimate goal of self-integration is necessarily multi-disciplinary. To be successful, that research must be grounded in the existing standards, technologies, and applications, yet build on evolving technologies such as the Semantic Web. Moreover, to be applicable, the research must be conducted jointly with all stakeholders including software vendors, manufacturers, standards organizations, and university researchers. With these complex research requirements in mind, we have initiated development of a B2B Interoperability Testbed. This testbed will support four major activities:

• to help the Supply Chain Management (SCM) community better understand the capabilities of Semantic Web and other formal methods for self-integrating systems,
• to affect the development of Semantic Web technologies and standards while representing needs of the SCM community,
• to facilitate introduction and acceptance of those standards by the SCM community, and
• to demonstrate the potential of the new generation of self-integrating software applications.

The testbed will provide an infrastructure for interaction among manufacturing companies, software vendors, and standards organizations. The manufacturing companies will provide interaction scenarios (as a basis for supply chain integration testing) and requirements describing context for integration testing (e.g., messaging protocol specification). We will capture these scenarios and requirements to drive testbed development and make available a scenario repository. Both the manufacturing companies and software vendors will provide and operate nodes of the distributed testbed as no central authority will exist over the testbed operation. The customers and vendors will come together on an as-needed basis to assess, analyze, measure, and demonstrate on-demand integration of software applications that are used to operate supply chains. We will provide guidance for coordination of the interactions among the different nodes, develop conformance tests, provide test data, conduct integration tests, analyze these tests, and report results.

The current and future projects in self-integration constitute fundamental activities within the testbed. The context in which these projects get created depends on the expressed needs of the SCM community (i.e., technology pull) and/or on the perceived opportunity for a particular Semantic Web technology (i.e., technology push). Each project will have partners from the academic, business, and standards communities and will conclude with a demonstration. The testbed will provide a mechanism for the transfer of the results of research projects into commercial products that address real, supply-chain-integration problems using the Semantic Web.

Anticipated Impacts:

A successful program will have the following:

Industry impact

• Reduction in integration/product costs
• Faster time to market
• Broader access to customers/suppliers
• Increased focus on core business

NIST impact

• New class of standards and ways of testing conformance
• Fewer interface specifications means fewer standards committees
• Unified approach to measuring both physical and information objects

Standards Participation:

Foundation for Intelligent Physical Agents (FIPA), Geneva, Switzerland
FIPA is a quasi standards development organization. Their interest in software agents includes the manufacturing domain and, specifically, Manufacturing Execution Systems. The NIST Framework project made initial contacts with FIPA in fiscal year 1998 and co-organized the first meeting of the Special Interest Group (SIG) on Product Design and Manufacturing at FIPA. (The Framework project was a project to establish a framework of standards and specifications for manufacturing software systems.) FIPA is a quasi standards development organization. Their interest in software agents includes the manufacturing domain and, specifically, Manufacturing Execution Systems.

UN/CEFACT ebTWG
Editor of the Catalog of Common Business Processes. The recently formed ebTWG (electronic business Transition Working Group), which is a continuation of ebXML work within UN/CEFACT, has initiated the Catalog of Common Business Processes project with the goal to define a useful taxonomy and methodology for mapping industry business processes onto the taxonomy.

OMG Manufacturing Domain Task Force, Several Committees
MEL participates in OMG's Manufacturing Domain Task Force (Mfg DTF). This organization is responsible for setting the overarching strategy for the working groups that comprise the task force. Specific activities include:

• OMG MfgDTF, PDM Working Group - This working group is responsible for issuing request for information (RFIs), Request for Proposals (RFP) and developing standards for PDM component interface definitions.
• OMG Web Ontology - Revived Ontology working group at OMG. Developed mission, planned deliverables, and actors identified to produce those deliverables. One of the key deliverables is one or more mappings from OMG modelware -- UML, MOF, and XMI -- to DAML and OWL
• Testability Work Group - Incorporated testability enhancements into manufacturing-related standards and validate their IDL on a regular basis.

W3C Web Ontology Working Group
Provided an extensive review of the proposal document for a UML Presentation Syntax for OWL.

Open Applications Group
OAG is a consortium of enterprise applications software vendors (i.e., ERP vendors) who have created common standards for the integration of enterprise business applications. This group produced the Open Applications Group Integration Specification (OAGIS), which is a requirements definition of information to be exchanged among ERP systems. The requirements specified in OAGIS are very relevant to the OMG ERP effort.

Send questions or comments to Webmaster.
Date Created: December 9, 2002
Last Modified: July 10, 2003