Bernus,P., Nemes,L., Morris,R. ``Possibilities an Limitations of Reusing Enterprise Moels,'' plenary paper, 2n IFIP/IFAC/IFORS Workshop on Intellingent Manufacturing Systems (IMS'94), P.Kopacek (E) OPWZ, Vienna, Austria (June 1994) pp11-16. [extene version appeare as: P.Bernus, L.Nemes, B.Morris, "The meaning of an enterprise moel" in Moelling an Methoologies for Enterprise Integration, P.Bernus, L. Nemes (Es), Chapman an Hall, Lonon (1996) pp 183-200]
Moels play various roles in the life cycle of the enterprise, an only some of them are executable. From those which are, only a part is machine executable. Moels also ten to be big, har to maintain, complex to analyse, an even worse (Nemes-Bernus, 1994) { from the incompleteness of enterprise moels it follows that there is no formal way of eciing whether the moel is only incomplete or it is inconsistent. Thus to guarantee consistency (an completeness as it will later be ene), users nee organisational measures (special functions built into the enterprise engineering process to establish moel consistency), or institutional guarantees (use of stanar moels). 1.3 The structure of this article Section 2 investigates where the meaning is in Enterprise Moels. The analysis reveals that enterprise moels are rst of all a means of communication between people to ensure a common unerstaning of the present or planne enterprise. Only a part of EM-s are use to control business processes, an even that executable part is mae possible only because of the rst essential function: meiating common unerstaning between those who esign, engineer an operate the enterprise. From the same analysis completeness criterion is erive, applicable to enterprise moels. Section 3 raws the consequences on the enterprise engineering process, presents limitations of reuse, an suggests ways to ensure successful reuse. Section 4 eals with the implications to enterprise engineering tools an environments (e.g. CASE tools). 2 WHERE IS THE MEANING IN EM? The meaning of a moel can be ene in more then one legitimate way, thus substantial confusion arises when one talks about the meaning of a moel. Three important meanings are investigate: the moel theoretic, the enotational, an the situate meaning. It will be obvious to the reaer that all of these contribute to the perception of what a moel means, thus these theories of meaning are complementing one another. 2.1 Moel theoretic meaning: An illustration In moel theory meaning is a mathematical structure representing an interplay ofsyntax an semantics. This subsection gives an illustration of moel theory (skipping some technical etails with reference to the literature (Lloy, 1994)). Figure 1 presents an Extene Entity Relationship iagram typically use when enterprise ata have to be moelle. The gure intens to say that: name supplier part specs elivery, quantity, contract-no (1,n) (1,1) specs-oc-number Q Q Q Q Q (1,n) supplies Q part in QQ esign QQ part esign rawing-number Fig. 1. ER Schema Describing a Sample Universe of Discourse \Part specications can be met by various esigns offere bypart suppliers at various elivery times in various quantities. Such elivery contracts, then are ientie bythepart specication (its ocument number), the name of the supplier, an the rawing number of the part. Furthermore to ientify a esign one has to know the supplier (because the rawing number is unique only within the numbering system of a given supplier). If a supplier supplie apart at a given time in a given quantity then that is registere inacontract with a unique contract number. A given part specication gives rise to a unique contract." Such an Entity Relationship iagram is expresse in a formal graphical language. To emphasize this point it is possible to translate what the ER iagram says into a list of logical propositions. Figure 2 shows part of such a translation (expresse say in Prolog). (Similar translation can be one into the atabase programming language SQL.) 2. If this formal representation is implemente as a Prolog atabase (as on Fig.2), an the atabase is lle with facts about iniviual entities, then queries from this atabase shoul return answers that correspon to reality. In other wors, the atabase is in fact a theory that escribes reality, an this theory has the preictive power to tell what woul be the outcome of any question if teste on reality instea of testing it on the atabase! However, since Fig. 3 represents the same ER iagram, it represents the same atabase as well. This means that the theories emboie in the two iagrams have the same preictive power they moel reality to the same epth. The fact that entity types have meaningful names on Fig.1 (as oppose to Fig.2) oes not inuence the behaviour of the atabase erive from it. In Moel theory it is customary to ene the meaning of a theory (i.e., here the meaning of the ER iagram) as the simplest, minimal moel of the theory. It follows that the rst ER iagram oes not actually 2 For the simplicity of emonstration the translation supposes that entities have immutable object ientiers. E.g. the entity set supplier has entities sup-1, sup-2, etc with name attributes Intel, DEC, etc... 2
Sample Translation of the ER Data Moel: entity types (entity sets): et(t) { t is an entity set entities: ei(t,o) -- entity o is an instance of type t relationship types: rt(t) -- tisarelationship type relationships: ri(t,r) -- relationship r is of type t attributes of entities/r'ships: at(t,a,) -- type t has attribute a with omain subtyping: st(sup,sub) -- sup is a supertype oftype sub key attributes: k(t,l) -- attribute list l is key of type t attributes of entities/r'ships: e(o,a,v) -- entity o has attribute a with value v etc... Axioms (integrity constraints): \entities of a type t have a unique key value" = 8t9k(8x1 x2 ei(t x1) ^ ei(t x2) ^ (x1 = x2 $ e(x1 k v) ^ e(x2 k v))) etc... Inference rules: \inheritance of attributes"= 8 T sub T sup Ast(T sup T sub) at(t sup A)! at(t sub A): etc... Sample Translation of the ER Schema of Fig.1: et(supplier). ei(supplier,sup-1). rt(suppliespartindesign). ri(suppliespartindesign,r1). at(supplier,name). k(partdesign,[rawingnumber,name]). e(sup-1,name,intel). etc... Fig. 2. Translation of an ER Schema Into Logical Propositions an Rules. N Fig. 3. SP PS SDN (1,n) (1,1) D,Q,C QQ Q Q Q (1,n) S Q QQ QQ P D ER Schema \Equivalent" to that of Fig.1 capture more of reality than the secon. This is because the structure an interpretation of both are the same. Even though in Fig. 1 \part specs" intuitively refers to a Part Specication, the atabase can not tell anything more about the nature of Part Specication than an (ientical) atabase erive from Fig.3! It also follows, that all systems onto which Fig. 1 can be mappe are equally escribe by Fig. 3. Formal software specications have the same nature the meaning of the specication is not eecte if the names of symbols use in the specications are change. Such a "moel" is complete, in the sense that no ambiguous interpretation is possible provie the formal specication language is chosen with care. It is "only" to be ensure that the unique moel so attribute to the specication (with suitable 1:1 mapping of language tokens to real worl entities an relationships) is the same as the real wor moel an that the suitable renaming is an intene renaming. If one compares the use of the two "moels" in Fig 1 an Fig 3, it is obvious that some other type of meaning shoul also be attache to our enterprise moels. Note here that one of the reasons why meaningful names are neee is because users want guarantees that their universe of iscourse (i.e. the relevant part of reality) is a moel of the theory emboie in the "moel" (here the ER iagram). 2.2 Denotational meaning As illustrate in Fig. 1, terms of a moel refer to enotations which are concrete or mental entities or relationships. Symbols of a language have enotations through language conventions an therefore the enotations are common to a given language community. In the case of Enterprise Moels, this community shoul be that of a profession in which the enterprise oes business. Encyclopaeias, ictionaries, technical textbooks oer escriptions of these stanar meanings. There are two ambiguities that spoil the simple enotational theory of meaning. First, enotations are often context epenent, thus a useful theory of meaning must take into account the contextual elements. Secon, enotations are common to a language community rather then a language alone the looser the connections in that community the less one can trust the enotational meaning. It is unfortunate that the community of persons who interact with Enterprise Moels o not form a single language community, so enterprise engineering processes nee built-in assurances to lter out the consequences ue to false interpretations which arise from incorrect enotations. E.g., the iterative proceure of author/reviewer cycle of IDEF-0 is intersperse with gures for \emonstration only" to establish this common interpretation of terminal symbols (Marca- McGowan, 1988). These \emonstrations only" are the brige between language communities. Demonstrations can be pictures, rawings, movie frames, or even other moels. The reaer might conclue that the moel theoretic meaning together with the enotational meaning of EMs suciently explains the nature of meaning communicate through EMs. By the combination of the two above theories of meaning a vast set of moel theoretically possible moels can be exclue from being possible interpretations. Users are not very intereste in which other moels may exist for the same theory if the symbols in the iagram were groune in an unintene way.when using EMs they implicitly assume that this mappingback to reality of meaningful symbols remains as intene. This in turn allows EM-s to be less etaile than they shoul be if only the moel theoretic meaning were use. As the next subsection shows even less etail is enough tolerable without jeoparising uniform interpretation. 3
2.3 Situate Meaning 2.3.1 Eciency an Completeness. To create an all encompassing moel of an enterprise is seemingly a aunting job, not only because of the complexity of the task, but more importantly, because the ever changing \organic" nature of business makes the enterprise a moving target for the moeller. Any moelling tasks that is to be one in a particular enterprise integration project must be one in a short perio of time. It is thus imperative to capture enterprise moels in a manner as ecient as possible. To achieve this eciency it is necessary to ene what the necessary level of completeness for enterprise moels is, because this criterion has major inuence to eciency. Furthermore, EM is not practical if the size of moels to be encountere by any one person is beyon the capacity (time, prerequisite knowlege, tool support etc) of the iniviual. Eciency an completeness are ene here in the context of the use of these moels: Def.1. Eciency An enterprise moel is ecient if it conveys the intene meaning concisely between the parties who prouce or use the moel. 3 Def.2 Completeness of enterprise moels An enterprise moel is complete relative to the processes using the moel if the processes can create (an behave accoring to) the intene interpretation of the moel. Three important consequences of the enitions are If there is no process that uses a moel, there is no nee for that moel. For example, the quest for an integrate corporate atabase schema of the 1980's (Smith et al, 1981) was awe because there was no nee for the complete schema. Only those parts of atabase schemata { the feerate schemata { shoul be prouce that integrate ata neee by some meaningful business process (Shet-Larson, 1990). It is not necessary that the EM be a true moel (or even a theory) at all! The only requirement is that the EM constrain its user in such away that only the intene interpretation is create in, or by the user. This point is explaine in etail in section 2.3.2 Notice, that even the interpretation of the moel nee not be fully mae explicit by any one user { the only pragmatic requirement is that when a "moel" is use the user shoul be able to create that part of the intene interpretation which is pragmatically useful for the user's actions. It is thus apparent that enterprise moels nee to maintain the eciency of natural language (incluing written an spoken wor, gures, etc) { while aing accuracy an formality. Eciency is a key factor. How is this possible? Semantic theories of natural language (Barwise, 1988) an (Barwise an Perry 1983) can be use to ene a thir meaning of enterprise moels which explain eciency. 3 Note the use of the wor \conveys" instea of \contains." 2.3.2 Situate Theory of Meaning. Below is a short exposition of the situate theory of meaning, or situation semantics for short. Situation semantics analyses meaning in the context of language use, such as a conversation or a cooperative action which involves language use. Utterances an escribe situations. Participants of aconversation pass on pieces of information to other participants in form of spoken or written wor, rawing, or other accepte moality of communication. Any such piece of information is calle an \utterance." Utterances are prouce with the intention e to a to, or moify, the recipient's moel of a topic, or \escribe situation." Interpretation is the process by which the recipient uses the utterance to buil or moify its own moel of the escribe situation. Since the recipient normally has an extensive set of moels available about a range of situations (past experience, common moels of a el of expertise), the recipient nees only a small set of utterances to buil the intene moel of the escribe situation. Atleast the set of utterances can be small compare to the moel. Utterance situation. Any utterance is uttere in an \utterance situation," which inclues the speaker, the listener, some agreement about the goal of the conversation, an possibly other circumstantial elements. This utterance situation is either irectly perceive by the participants or is of some stanar form (e.g. the proucer of utterances can anticipate the situation in which the recipient will interpret the utterances). Note that the same utterance may be interprete in very ierent ways if the utterance situation changes. Meaning as a relation. The situate meaning of an utterance is the relationship that the utterance establishes between the utterance situation an the escribe situation. This relationship enables the recipient to restrict the set of possible escribe situations an thus helps buil the internal moel of the escribe situation. The conversation is successful if the recipient is able to re-create the intene interpretation. It follows that there are three elements which can be controlle for the act of communication to succee: Utterances Utterance situation Describe situation. 2.3.3 Situate Meaning of Enterprise Moels. An enterprise moel will be thought ofasasetofutterances intene to convey in a precise an ecient manner some information about the enterprise. The goal of enterprise moelling is to achieve a target situation (some new, improve state of the enterprise). This target situation is the escribe situation of EMs. If that escribe situation is constraine from the outset then the eciency of EM is improve. The interpretation of an EM is embee in \enterprise engineering situations." This inclues experts, 4
reference materials (previous knowlege of paraigmatic cases, stanar moels, experience), an most importantly the methoology which is followe in the process of enterprise engineering. When moelling-experts prouce a moel an communicate it to some recipient group, the intention of this communication is alreay xe, the participants are known, an the suppose prerequisite knowlege of the participants may also be ene. All of the above elements form part of the utterance situation in which the EM is to be interprete. EMs, as a collection of utterances, thus contain information only inasmuch as they constrain the user suciently so that only the intene interpretation is create. Enterprise moels o not necessarily contain the information that they are usually suppose to carry! Def.3 The situate meaning of enterprise moels is the relationship between the situation in which the EM is communicate an the situation about which the EM is stating something. This pragmatic treatment of meaning allows for a hugely increase eciency in EM communication, an may be implemente as a new way of using enterprise moels { provie one can have goo control over the \utterance situations" an over the initial state of the \escribe situation." The same treatment also allows for completeness to be ene relative to the process of use of EMs rather then completeness as an intrinsic property of EMs. Denition 2 gave completeness as a pragmatic property ofenterprise moels. It is now clear that this pragmatic completeness coincies with the EM unambiguously an suciently carrying its situate meanings. In contrast with ef. 2, an absolute measure of completeness is the extent towhichenterprise moels are theories (in the sense of moel theory) { together with some enotational assignment { as oppose to being utterances allowing the recipient to create these theories. One woul think that if an EM is complete in the absolute sense then it is also complete in the relativistic sense. However, this is not so: the recipient of the moel may not possess the requisite abilities, or tools to correctly interpret a moel which may be juge complete in the absolute sense by an omniscient external observer. 3 LIMITATIONS AND POS- SIBILITIES OF REUSE 3.1 Possibilities of reuse Clearly the important conition of successful moel reuse is that moels be pragmatically complete (Def.2). Figure 4 shows at a glance the factors which together form the enterprise engineering situation. The \escribe situation" here is the interpretation particular project moel enterprise. percieve interpretation reference of particular. enterprise moel architecture. particular?? enterprise moel. - partial enterprise moel. - encyclopaeic references (multimoal reference atabase) -. moelling an 6 simulation tools b b 6 enterprise engineer with - role, motivations - experience with type @@ of system - experience with moelling language - experience with reference architecture - isciplinary backgroun knowlege - in-house (particluar) system knowlege? any enterprise engineering activity using an EM Fig. 4. Factors eecting the pragmatic interpretation of enterprise moels - any action or communication base on percieve interpretation of EM-s as perceive by the enterprise engineer. 4 For an EM to be (re)usable as intene, it is thus necessary to specify what is suppose about the use of the moel in terms of all these factors (as liste on Fig.4). These factors of successful reuse are to be reprouce in the "reusing" process: Qualities of the enterprise engineer Reference to the type of enterprise engineering situation in which the EM is to be use. (E.g. by reference to an enterprise reference architecture. 5 Ability to explicitly view the current moel of the enterprise engineering process. Quick access by theenterprise engineer to the wie range of reference material which may have been use in the prouction of the EM. Links in the esign atabase between the partial an particular moels an the enterprise engineering process. Capturing of the esign history. The complexity of the enterprise engineering process (an the moels prouce in it) can be signicantly reuce by stanarising many of the above components, as: Enterprise Reference Architectures Ontologies / Partial Moels. 4 The term enterprise engineer is use as a shorthan in place of naming all those persons who are involve in interpreting enterprise moels. 5 Enterprise Reference Architectures are moels, accompanie by methoologies, of the enterprise engineering process { encompassing the entire life cycle of the enterprise. 5
3.2 Limitations of reuse The lack of the same factors which enable the intene interpretation to take place can limit the possibility of reuse. To prevent problems it is necessary to review the prerequisites of successful interpretation at the planning stage of any enterprise engineering process. The lack of aherence to an enterprise reference architecture can have a signicant negative eect on the success of reuse. When moels are prouce they ten to be precise with respect to the intene use an ignore etails not necessary for that use. Failure to unerstan the original intene usage is a misuse of EMs an is a major cause of misinterpretation. The lack of prompt access to aequate encyclopeic reference material. Both the access an the promptness of this access are necessary since failure to use the reference when it woul be necessary introuces an unue (an maybe unnotice) backtracking point to the esign process. CASE tools of a new type (see section 4) can alleviate the enotational an some of the situational misinterpretations, while simulation tools can help investigate the implicit properties of (executable) formal specications. Below a set of functions are outline that enterprise engineering CASE tools shoul perform to ensure that the enterprise engineering process puts each participant in a position (situation) where correct interpretation is possible. 4 IMPLICATIONS TO ENTERPRISE ENGINEERING TOOLS AND ENVIRONMENTS Traitional CASE tools allow the esigner to create a moel in a chosen moelling language. They o not help, however, other esigners to unerstan that moel. Below is a shortlist of requirements that enterprise engineering CASE tools nee to satisfy. Only those factors are liste here which are not usually part of a state-of-the-art CASE tool. Oer links to reference material an cross references to other EMs Be permissive { allow multiple moelling languages Translate between various representations of the same EM through algorithms or common reference Make the enterprise engineering process explicit an up-to-ate (reference moels an particular moel) EMs are to be communicate between people an mutual unerstaning is to be an observable occasion in this process (support negotiation, iscussion, an in general, cooperative group work) 6. Ability to navigate in the EM via queries an links Ability to iscover implicit properties by simulation The "Intelligent CAD" community has argue that both the representation an automatic control of esign ata an esign processes is neee to support esign. Inee enterprise engineering tools with the above qualities will have a kin of intelligence in this respect namely that they enhance the intelligence of their users although not necessarily isplaing some "inherent" form of intelligence. Networke information iscovery tools seem very appropriate for the provision of the extensive reference functions with intelligent querying facilities to locate the references. 5 CONCLUSION Reasons of incompleteness of enterprise moels were analyse an a enition of pragmatic completeness was given which can, an shoul, be achieve in enterprise moelling. It was emonstrate how enterprise moels carry meaning. This resulte in requirements for the enterprise engineering process, which { if not met { can limit the viability of the process. The analysis of the same factors resulte in requirements for improve enterprise engineering CASE tools. REFERENCES Barwise,J. (1988) \The Situation in Logic," Stanfor,USA, CSLI,1988 Barwise,J., Perry,J. (1983) \Situations an Attitues," Cambrige,MA, MIT Press. Dorval,B., (1990) \Conversational organization an its evelopment," (B.Dorval e.), Ablex, Norwoo. Lloy,J.W. \Founations of Logic Programming," Springer, Berlin,1984 Marca,D.A., McGowan,C.L (1988) \SADT," McGraw Hill, New York Nemes, L., Bernus, P. (1984) \An Incomplete Manufacturing Moel Nees Matching Design Tools," Proc. 16th CIRP Int. Sem. on Manuf. Sys. Tokyo, pp.26-43 Sheth,A.P., Larson,J.A. (1990), \Feerate Database Systems for Managing Distribute, Heterogeneous, an Autonomous Databases," ACM Comp. Surv. 223, 183-236 Smith,J.M., et al (1981), \ Multibase { Integrating Heterogenous Distribute Database Systems," Proc. AFIPS, 50,5 487-499 T.J. Williams P et al, (1993) \Architectures for Integrating Manufacturing Activities an Enterprises," Prepr. IFAC'93 W.C., Syney, X, 273-283 6 Theories of conversation (Dorval, 1990) emonstrate that to achieve mutual unerstaning the participants nee to have an agreement on what constitutes the present esign situation, an have access to the same common references for enotational (e.g. experimental) purposes. 6