This workshop will seek to bring together three diverse groups of people with the explicit goal of developing shared methodologies for building and using large expressive models.
1. There are practitioners in industry who are attempting to build models of large, complex and evolving systems. Traditional industry practice emphasizes disciplined approaches, formal development methodologies, and stepwise validation in the development process. This practice is often realized by the user of "simpler" model representations such as, E-R, IDEF, Data Flow or OO.
2. There are AI practitioners who define and use rich representational frameworks. Since the work pushes the state-of-the-art, AI models are small and focused on particular aspects of what needs to be modeled. Recently, however, some attempts have been made to build models in the large using richer representations.
3. Those with experience in knowledge engineering methods tend to focus on methodologies of acquiring and validating knowledge. Such methods tend to be task-specific and domain-focused. Here too, the issue of scaling up to large, complex and evolving models begun to emerge.
This workshop follows closely two successful workshops: 1991 AAAI Workshop on AI in Enterprise Integration and the First Annual Conference on EI Modeling Technology (ICEIMT). The proposed workshop will attempt to focus on blending the practical experience base of enterprise modelers with richer methods of knowledge representation and knowledge acquisition for a fertile interchange of information.
Models in the large typically tend to be
o Enterprise Model- i.e. how an enterprise is organized, how it functions, including material flow, information flow, financial flow, decision making and goal setting etc.. An Enterprise Model could be that of an "as-is" enterprise or a "to-be" one. The term "enterprise" in this context applies to an organization that conducts business by providing products or services; hence it could be a company, university, department, division, partnership or whatever.
o Models for supporting concurrent design and engineering.
o Design knowledge capture in large projects (e.g. the NASA Space Station program.)
Models that have been attempted vary in richness from data models, information flow models to full semantic or logic models. The richness of models determines the multitude of uses to which a model can be put. Hence the development of semantically rich models that permit reasoning over the model structure and contents are of great interest and importance now. AI, knowledge representation and knowledge engineering methods have much to contribute to this effort.
Topics considered include:
ENTERPRISE MODELING Formalism easy to learn, teach and use Methods for dealing with scale up Dealing with consensus reality Dealing with how organization changes while modeling is in progress Learning and adaptation in organizations Open system hypothesis Capturing Mission, Strategies, Goals and Plans Reasoning about Policies and Procedures Performance indicators at unit and aggregate levels Product & Process integration Concurrent Design and Engineering Business Process Re-engineering Business process simulation Justification of the expense and effort of modeling to executive management
MODEL REPRESENTATION Frame, logic-based, and Description logics Semantics of key constructs like part/wholes, flows and teleology Actor systems DAI techniques including negotiation Reasoning Methods: formal and heuristic Three Schema view -Use of separate representation for communication with content experts, discussion among the modelers, and for computer representation Relation to Object Oriented Models, Semantic Data Models etc.
KNOWLEDGE ENGINEERING METHODOLOGIES Knowledge Engineering -Sources of models: interviews, manuals, observations -Modeling "consensus reality" -Model integration and validation -Model consistency checking -How to tailor the "purpose" of modeling into the "methods" Information Engineering Methodology -Expansion of this to technical domains -OO extensions to IE Relation to OOA/OOD/OBA/SADT etc.
Significance: In the corporate computing circles the topic of Enterprise Integration and Enterprise Modeling are of great interest and multiple million dollar efforts are beginning. However, the formalisms chosen for many of the modeling methods utilize simple data flow graphs and hierarchical system views of the world. Knowledge representation and Knowledge engineering techniques of the past 25 years have much to contribute to these efforts. Yet, no discipline and methodology appears to be emerging from the AI efforts. It appears timely to capitalize on a wonderful opportunity to target AI techniques in the service of Enterprise Modeling.
Format: The workshop is planned for a day and a half with presentations of papers and invited talks, open discussions and 2 panels. Several "joint" presentations will be given in intersection areas. About 40 people are expected to be invited. All participants will be selected based on expected contributions to the cross-fertilization among several topics, rather than strength or experience in one topic.
Submission requirements: Interested participants should send in a position paper of 4 to 5 pages by electronic mail (preferred) or hardcopy. This should cover thoughts on important and interesting areas of concern for "Modeling in the Large" with particular emphasis on disciplined methodologies. A brief statement of relevant background or experience of the author should be included.
Papers should be received by the workshop chair by March 12 1993. Notification of acceptance will by on April 2. Camera-ready copies are required by April 30.
Please send your submissions to N.S. Sridharan at the address below.
N.S. Sridharan ("Sri") Intel Corporation, MS-CH2-23 5000 W. Chandler Boulevard Chandler, AZ 85226 (602) 554 3324 (602) 554 7116 fax NSridharan@faois.intel.com
Organizing Committee:
Robert Filman Intellicorp
H. Firdman ("Eric") Pacific Bell
Neil Iscoe EDS Research
V. Jagannathan ("Juggy") CERC, West Virginia University
Jim Schmolze Tufts University
J. Tenenbaum ("Marty") Enterprise Integration Technologies
Gerry Williams Andersen Consulting