http://www-ksl-svc.stanford.edu:5915/FRAME-EDITOR/UID-21&sid=ANONYMOUS&user-id=ALIEN Subject: Explicit scoping criteria for the SUO Date: Thu, 10 Aug 2000 11:49:30 -0700 (PDT) From: mfu@redwood.rt.cs.boeing.com (Michael Uschold) To: standard-upper-ontology@ieee.org FURTHER THOUGHTS on SCOPING Perhaps our experience in scoping the Enterprise Ontology (EO) is relevant here. It may be conceived as an 'upper level' ontology for enterprise modeling. It was intended to: "provide the foundation for ontologies of much larger size and more specific scope.", just as the SUO is, except that it was more limited to start with. The same is true, for PIF/PSL in the domain of processes. But I think the difficulties of deciding what to include and what to exclude are similar. To determine what terms went in the EO: " The ultimate criterion is the judgement of what concepts are likely to be sufficiently important to the Enterprise Project and be capable of a common agreement on their meaning. [Uschold et al 98]" See 'appendix' for full text of the scoping section in this paper. The enterprise project goals include the capability to build enterprise models, and that a key goal of the ontology was to provide a basis for inter-operability among different systems. Thus, what the above criterion amounted to was: a term should be included IFF it was going to be important to enable customers to build their required models. Now, what I realize here is that this does not provide guidance as to when a term is too domain-specific to warrant inclusion. We had no explicit guidelines, but in retrospect, I believe our implicit (i.e retro-fitted by me here :-) criteria were something like the following (Martin King please add to this if possible): A term should be included if: It refers to a concept that is common among most or all enterprises. Thus it would include things like SALE or MANAGER, but not DORITOS, or AEROSPACE which are relevant only to certain enterprises. If a customer paid good money for the EO, and s/he was told that it covered all the basic concepts in the enterprise modeling domain, then that customer would expect this term to be in the ontology. Conversely, a term should be excluded if it is likely to apply to only a small number of enterprises or that our hypothetical customer would never dream of complaining that this term was not in the core EO. How to apply this to the SUO? It's really a question of determining what is domain specific vs domain independent. Perhaps some of the PIF/PSL team could report on what criteria they used for scoping. I seem to recall that the PIF and PSL teams had slightly different approaches -- one had an engineering/implementatin focus, the other had a user focus. Michael Gruninger, can you elaborate briefly on this? CYC is a more relevant case for determining scope. They have a customer focus, I belive. Fritz may perhaps help enlighten us? I think this project would benefit if we could agree on some up-front criteria about what should be in and what should not be in the SUO. Michael Gruninger uses the competency question approach, to great effect. That may not be practical for th SOU, but I think we will benefit from some criteria because they will keep us focussed, and always provide a fall-back when disagreements arise about what to include/exclude. APPENDIX Full text of the scoping section of our paper on the EO. [Uschold et al 98] M. Uschold, M. King, S. Moralee and Y. Zorgios; The Enterprise Ontology Knowledge Engineering Review 13(1), 1998") SCOPE Considerable time and effort has been devoted to deciding the scope and boundaries for the Ontology. We began by brainstorming to identify as many potentially important concepts as possible. This produced a totally unstructured list of words and phrases corresponding to a wide variety of concepts relevant to Enterprises. These were then grouped into various more or less distinct work areas such that there was more similarity in meaning and a need to refer to terms within an area than between different areas (e.g. Activity, Marketing, Organisation). Within each work area, the terms were assigned priorities indicating the importance of including them in the Ontology. At this point many terms were discarded and duplicates (i.e. nearly synonymous terms) were removed. These work areas were dealt with one by one. For each concept, terms were chosen, and definitions given. The original work areas evolved somewhat, as new terms were added, and others removed or moved to other areas. Eventually, these work areas became the major structuring element for the Ontology and is reflected in the major sections of this document. Within each work area, various important questions were addressed. What basic or core concepts are required? What mix of terms having a wide or general meaning and terms having a narrow or specific meaning are required? Many factors influenced the choice of terms in the Ontology. The ultimate criterion is the judgement of what concepts are likely to be sufficiently important to the Enterprise Project and be capable of a common agreement on their meaning. Many words in common use in enterprise management have been judged to have no sufficiently widely recognised or acceptable meaning to be included in the Ontology. This does not mean they cannot be used in the project. It does mean that the meaning of such words in the context of their use will have to be related to the terms in the Ontology all of whose meanings are shared. This document attempts to give guidance on how this can be done where a potential need for this has been recognised.