1 STPO: The synthesis of a TOVE Persons Ontology Chris Partridge Technical Report 24/02, ISIB Padova, Italy, December 2002 Questo lavoro è stato condotto nell'ambito dell'attività del gruppo di ricerca “Modellazione concettuale e Ingegneria della Conoscenza” del ISIB-CNR ISIB-CNR Corso Stati Uniti 4 I-35127 PADOVA (PD) e-mail: mbox@ladseb.pd.cnr.it fax: +39 049-829.5763 tel: +39 049-829.5702 -- 1 of 108 -- 2 -- 2 of 108 -- The Synthesis of a TOVE Persons Ontology Page 1 Copyright © 2003 - Chris Partridge. All rights reserved. Chris Partridge1 & 2 1 BORO Program, England partridge@BOROProgram.org http://www.BOROProgram.org/ 2 National Research Council, LADSEB-CNR, Italy partridge@ladseb.pd.cnr.it http://www.ladseb.pd.cnr.it/infor/ontology/BusinessObjectsOntology.html Abstract: This is a report of the results of the Synthesised TOVE Persons Ontology (STPO) project. This project’s goal was a synthesis of a Persons Ontology from the TOronto Virtual Enterprise (TOVE) project’s Organisation Ontology. The report is both an introduction to the interim ontology developed by the project and also a summary of its development. 1 ‘The STPO Final Report’ This is a report of the results of the STPO (Synthesised TOVE Persons Ontology) Sub-project – a step of the CEO (Core Enterprise Ontology) Project. It is intended as both an introduction to the interim ontology developed by the project and also a summary of its development. The body of the report focuses on the work of the STPO. Background information on the CEO and its approach (as implemented in the STPO) can be found in the report The CEO Project: An Introduction (Partridge 2002a). The STPO is the first step of the CEO project whose eventual aim is the construction of an ‘industrial strength’ ontology to be used as a tool by enterprises to significantly improve the semantic aspects of their information systems. This first step is a synthesis of a Persons Ontology from TOronto Virtual Enterprise (TOVE) project’s Organisation Ontology1 - which is called the Synthesised TOVE Persons Ontological Model (STPOM). 1.1 The synthesis stage of the CEO project STPO is only a small first step in the overall CEO project – it is just the first step in the first ‘synthesis’ stage, as shown in Figure 1 below. The synthesis is a kind of informed, intelligent re-construction or recovery on the basis of a sound framework ontology2 . It has as its goal the harvesting of insights from current ‘state of the art’ enterprise ontologies and their synthesis into a single coherent 1 The TOVE project’s aim is to produce an Enterprise Model of which the Organisation Ontology is part. They have published a number of versions of this - the particular version that is considered here is Fox, et al. (1996) An Organisation Ontology for Enterprise Modelling. 2 And so does not fall neatly into one of the usual categories; for example, those of integration, merge and use in Gomez-Perez, et al. (1999) Some Issues on Ontology Integration, that assume an underlying homogeneity among the ontologies. -- 3 of 108 -- The Synthesis of a TOVE Persons Ontology Page 2 Copyright © 2003 - Chris Partridge. All rights reserved. whole. This will then be used as the foundation for the development of an industrial strength CEO. The planning of the synthesis stage started with an informal review of what was available. This indicated that the ‘state of the art’ is immature, in particular that: • there are not many enterprise ontologies (though there are many resources from which these could be mined), and • those that exist have not yet reached ‘industrial strength’ as ontologies for semantic interoperability of operational enterprise systems. This second point is one of the reasons why the STPO needed to undertake a synthesis rather than a merge/integration. This should become clearer as this analysis of TOVE proceeds. The review selected the following ontologies for synthesis: • TOronto Virtual Enterprise - TOVE (Fox, Barbuceanu et al. 1996) (Fox, Chionglo et al. 1993) (TOVE:http), • AIAI’s Enterprise Ontology - EO (Uschold, King et al. 1997) (Uschold, King et al. 1998) (EO:http), • Cycorp’s Cyc® Knowledge Base – CYC (Lenat and Guha 1990) (CYC:http), and • W.H. Inmon’s Data Model Resource Book - DMRB 3 (Inmon, Silverston et al. 1997) For planning purposes the review also made a rough intuitive guess at what the major core enterprise categories might be – sufficient to act as a basis for scoping the synthesis sub-projects. It proposed these three major categories: • Person (AKA Party4 ), (where this includes both natural persons and organisations) who can enter into a • Transaction, which often include agreements which involve an • Asset. The synthesis project was sub-divided on the basis of the selected ontologies and the proposed major categories – as shown in Figure 1 below. The intersection of the TOVE ontology and the Person core category was selected as the scope for the first synthesis project. 3 In its own terms, this is a universal data model. However, from our perspective, it is in many respects an ontology. We considered having a number of commercial data models in the sample, but found that they were very similar – so there would be no real benefit. Inmon, et al. (1997) The data model resource book and Hay (1996) Data model patterns were neck and neck as the commercial data model representative. We selected Inmon (1997) as it seemed slightly more accessible. Note, a two volume revised edition of this has since appeared: Silverston (2001a) The data model resource book 1, Silverston (2001b) The data model resource book 2. 4 Within the data modelling community, Party is the common name for what we are calling Person here. -- 4 of 108 -- The Synthesis of a TOVE Persons Ontology Page 3 Copyright © 2003 - Chris Partridge. All rights reserved. EO CYC DMRB CEO TOVE Persons Synthesis Stage Development Stage R E V I E W STPO Overall CEO Project Transactions Assets Figure 1 – The synthesis stage of the CEO project 1.2 Collaboration In the initial stages, the STPO project collaborated with the Italian segment of the European IKF (Intelligent Knowledge Fusion) project 5 : in particular with the pilot IKF-LEX project. IKF/IF-LEX is lead by ELSAG BankLab SpA and its goal is to provide semi-automatic support for the comparison of banking supervision regulations. 1.3 Organisation of this paper This paper is organised into these three main parts: • Introductory: The first sections of the paper contain this introduction and a description of the context for the overall CEO project and this STPO project. They outline the goal and scope of the projects as well as the ontological approach taken. • Main: The body of the paper describes the synthesis of the STPO ontology from the TOVE ontology. This is done in three stages based upon the three major elements of the TOVE ontology. • Conclusions: The final sections of the paper discuss further work that needs to be done in the CEO project – both in terms of analysis and expanding the ontology – and the conclusion that can be drawn from the work done on the STPO project. 5 http://www3.eureka.be/Home/projectdb/PrjFormFrame.asp?pr_id=2235 -- 5 of 108 -- The Synthesis of a TOVE Persons Ontology Page 4 Copyright © 2003 - Chris Partridge. All rights reserved. 2 The CEO project In this section, the main points relevant to the STPO sub-project are briefly described. As noted earlier, further details of the CEO Project can be found in (Partridge 2002a). 2.1 The CEO’s goal, scope and prime deliverable The overall goal of the CEO project is to provide a tool for enterprises to significantly improve the semantic aspects of their information systems. The CEO is intended to be a Core Enterprise Ontology, which means its scope is the major core categories of the enterprise field. The CEO’s prime deliverable is an ontological model – which provides a semantic framework that can be used for many tasks, including the development, deployment and inter-operation of systems. Technical terms, such as ‘ontology’ and ‘ontological model’, are explained in (Partridge 2002a). 2.2 The STPO sub-project The STPO’s goal and scope have been refined from the CEO’s. 2.2.1 The STPO’s goal, scope and prime deliverable The goal of the STPO is the harvesting of insights from TOVE and their synthesis into a single coherent whole within an ontological model – the STPO-M. This will then be used as the basis for further synthesis projects. As noted earlier, the choice of the enterprise category Person gives us one boundary on STPO’s scope, TOVE gives us another. So the STPO’s focus is on the intersection of these – the person-relevant elements of TOVE. The STPO’s prime deliverable is an ontological model that builds upon TOVE’s insights to arrive at something closer to ‘industrial strength’. 2.3 The CEO’s requirements for a useful ontology From the CEO perspective, ontology is a technology and the ontological model a tool for semantic integration. There are a number of requirements that it needs to meet to fulfil its purpose. The main general requirements identified in (Partridge 2002a) are divided into these engineering-based ones: • Teachability • Consistent applicablility And these science-based ones: • Relevant precision and sufficient formality • Sufficient simplicity and relevant generality • Appropriate unity and explanation • Relevant fruitfulness -- 6 of 108 -- The Synthesis of a TOVE Persons Ontology Page 5 Copyright © 2003 - Chris Partridge. All rights reserved. • Relevant repeatability – re-usability 2.3.1 TOVE’s stated requirements TOVE lists some similar key requirements in Section 3 of (Fox, Barbuceanu et al. 1996), under the heading ‘Ontology Competence’. It is worthwhile to note the similarities and the differences. TOVE regards the most importance criteria to be what it calls competence. This asks: “How well does it support problem solving? That is, what questions can the representation answer or what tasks can it support?” In practice, it uses the easily tested ‘questions that need to be answered’ as a yardstick for measuring competence: though how to arrive at a comprehensive set of competency questions is a moot point. The CEO and TOVE are in some kind of broad agreement inasmuch as competence is similar to the notions of ‘relevant’ and ‘appropriate’ used in the CEO requirements. TOVE mentions generality asking “To what degree is the representation shared between diverse activities such as design and troubleshooting, or even design and marketing?” This implies that TOVE has a more modest goal in this area that the CEO – limiting their stated interest to term harmonisation across domains. TOVE mentions perspicuity, asking: “Is the representation easily understood by the users? Does the representation document itself?” These questions bear a weak resemblance to the explanation and usability CEO requirements. Interestingly, TOVE mentions minimality: “Is there a core set of ontological primitives that are partitionable or do they overlap in denotation? A minimal set of terms should be in the ontology.” This minimality requirement is related to the CEO’s requirement for simplicity and generality. TOVE also mentions transformability, extensibility, granularity and scalability. But as expressed, these seem to be properties of the representation rather than the ontology – and so not relevant to our enterprise. 2.4 The CEO’s Framework for the Synthesis The CEO uses a framework of three hierarchies (described in detail in (Partridge 2002a)) each generated by its hierarchical relation. These are: • Typonomy - generated by the instance-type relation, • Taxonomy - generated by the super-sub-type relation, and • Partonomy - generated by the whole-part relation The initial review identified that, in general, the sample ontologies did not take full account of these frameworks – this is particularly true for TOVE. This is one aspect of the immaturity of the current crop of ontologies. An important part of the analysis will be developing a model of these framework for the objects recovered from TOVE. -- 7 of 108 -- The Synthesis of a TOVE Persons Ontology Page 6 Copyright © 2003 - Chris Partridge. All rights reserved. 3 TOVE’s Person scope The notion of Person (Party) includes both people and organisations: it encompasses all the entities that can enter into contracts and so acquire rights and obligations. TOVE has a number of person-relevant elements. The initial STPO analysis identified that these three main elements6 (and their sub-elements and their inter-relationships): • Organisation (and Organisation-Unit), • Organisation Agent (with Individual-Agent and Group-Agent as ‘sub- classes’), • Organisation-Position, There is also a related main element that is not in scope, but is relevant to the analysis: • Organisation-Goal. These elements and their relations, which I shall call TOVE’s Organisation Ontology, is diagrammed below. A list of all the elements in the TOVE ontology is given in Appendix A. Organisation- Entity Organisation Organisation- Goal Individual- Agent Group- Agent Organisation- Agent Organisation- Unit org-unit member-of unit-member unit-goal org-goal Organisation- Position agent-position/ filled-by unit authority member org- membership Organisation Organisation-Agent Organisation- Position Figure 2 – TOVE’s Organisation Ontology To simplify things, this diagram (and the STPO analysis) does not distinguish between relations and their inverses. To aid traceability back to the original TOVE relations, both the relation and its inverse’s names are shown by a single diagrammed relation line. Unfortunately TOVE does not seem to include any obvious way of identifying the inverse of a relation. So the STPO has had to use its judgement, based upon context, to identify inverses. In some cases TOVE makes this more difficult than it need be by using quite different names. For example, Organisation_org- 6 TOVE talks of concepts – i.e. the representation rather than the represented. To minimise confusion I have used the term ‘element’ rather than ‘concept’. -- 8 of 108 -- The Synthesis of a TOVE Persons Ontology Page 7 Copyright © 2003 - Chris Partridge. All rights reserved. unit_Organisation-Unit and Organisation-Unit_member-of_Organisation seem to be inverses – despite the quite different senses of org-unit and member-of. 3.1 The notion of organisation Organisation is one of the main person-relevant elements within TOVE’s Ontology. TOVE, and the rest of the sample, use the term organisation in a sense common in enterprises, which takes it as a type of person. In other contexts, the sense is sometimes stretched to include what might more naturally be called a system (‘a regularly interacting or interdependent group of items forming a unified whole’) – taking in such things as animals (including humans), which are a functioning ‘organisation’ of their bodily parts. The STPO, like TOVE and the other ontologies, keeps to the enterprise sense that includes such things as7 : • a voluntary association of individuals for common ends; especially: an organized group working together or periodically meeting because of common interests, beliefs, or profession. • an enduring and cooperating social group whose members have developed organized patterns of relationships through interaction with one another. • a community, nation, or broad grouping of people having common traditions, institutions, and collective activities and interests. • a chartered commercial organization or medieval trade guild. • an association of persons for carrying on a commercial or industrial enterprise. These extracts all refer, in one way or another, to groups of people. However, one of the things the STPO analysis showed is that the enterprise notion of an organisation is not dependent upon this – and this independence needs to be recognised to account for examples such as single person companies and sole proprietorships. 3.2 TOVE’s organisation ontology’s intuitions The goal of the STPO is to extract from TOVE its insights, both into the formalisation and the explanation of the enterprise – and synthesise them into a coherent whole. The kind of insights TOVE has captured includes a number of basic intuitions about the nature of organisations, which most people regard as correct in at least some situations. The kinds of intuition are: • An organisation can and does perform activities. • An organisation can be composed of units and typically has members. • People can be members of organisations. 7 These are based upon extracts from a variety of dictionaries. -- 9 of 108 -- The Synthesis of a TOVE Persons Ontology Page 8 Copyright © 2003 - Chris Partridge. All rights reserved. • Some (maybe all) of the organisation’s activities are performed by their members in some form of collaboration. • An organisation typically places some constraints on the activities its members can perform on its behalf. • Organisations have positions. • People occupy positions in organisations. These need to be recognised and explained by any CEO and the synthesis of TOVE provides us with the opportunity to do this. TOVE provides us with an obvious order in which to proceed with the synthesis – its three main concepts: • Organisation (and Organisation-Unit) (Section 7.1) • Organisation-Agent (and Individual-Agent and Group-Agent) (Section 7.2), and • Organisation-Position (Section 7.4). 4 TOVE’s Organisation We can look at the section of TOVE covering Organisation as attempting to answer these two questions: • What kind of thing is an organisation? (Organisation and Organisation- Unit) • How is an organisation composed of other organisations? (org_unit and unit relations) 4.1 What kind of thing is an organisation? This first question is asking for, from a formal perspective, a taxonomy, particularly the super-types of Organisation. To get at TOVE’s taxonomy, the STPO had to look at its informal descriptions, as its formal ontology has little to say on the matter. Effectively there is no super-type of Organisation – the specified super-type, Organisation-Entity, is just a catchall for the entities in the ontology. And there are only two sub-types: Organisation and Organisation-Unit. 4.1.1 TOVE’s informal answer Its informal description in Section 7.1 – Organisation – points towards an answer to the question: ‘To begin, an organization consists of a set of Organisation-Agents (said to be members of the organisation), a set of Organisation-Units (recursive subcomponents having a structure similar to organisations) and an Organisation-Goal tree that specifies the goal (and its decomposition into sub-goals) the members try to achieve.’ This reflects some of the intuitions mentioned earlier. Organisations can ‘consist’ (in one sense) of members or ‘consist’ (presumably, given that these are -- 10 of 108 -- The Synthesis of a TOVE Persons Ontology Page 9 Copyright © 2003 - Chris Partridge. All rights reserved. represented by different relations in its model, in another sense) of other organisations (units) – which, in turn, can also have members and units. However, to be useful the senses of ‘consists’ need to be specified more precisely. Does either of the senses of consist imply set membership, mereological fusion or something else? It seems unintuitive that an organisation is the set of two sets and a tree, but if this is really what is claimed, then it needs to be explained. The informal description in Section 2.0 – What is an Organisation – is less intuitively obvious: ‘We consider an organisation to be a set of constraints on the activities performed by a set of collaborating agents.’ Organisations clearly have something to do with activities, but, intuitively, an organisation (such as IBM) is not a set of constraints. (Maybe the authors meant ‘to have’ rather than ‘to be’ a set of constraints.) However, this description does imply potentially useful identity conditions, which seem capable of differentiating one kind of limit 8 case – where two organisations have the same members9 . In this case, they could be different because they involved different activities (and so different constraints) or the same activities with different constraints. Presumably, if they involved the same activities with the same constraints, they would be the same organisation. As these examples show, one of the major difficulties of providing an ontological explanation is crafting something that does not have awkward (or even worse, inconsistent) implications. If we take TOVE’s description of organisations as sets (of constraints) seriously, this raises a number of difficult-to-answer questions about Organisation-Unit and its member relation – such as: • If an Organisation-Unit is similar to an Organisation (as TOVE implies by saying they have “a structure similar to organisations”), then is it also a ‘set of constraints’? • What is the member relation between Organisation (a set) and Organisation-Unit? Is it set-membership or something else? If Organisation-Unit is a ‘set of constraints’, then the relation would seem to be more like sub-set. Different descriptions are a useful mechanism for capturing different intuitions. But where there is more than one description – there is the possibility of inconsistency, as the two informal TOVE descriptions above show. Taken 8 Limit cases are useful in determining whether criteria are merely good ‘rules of thumb’ or always apply by pushing an aspect to a limit – in this case, pushing membership as a criterion for identity to the limit. 9 Discussed, for example, on pp.220-1 of Gilbert (1992) On social facts, where the groups are differentiated as being qua this and qua that. And in Simons (1987) Parts p.168 4.9 Parts of Groups “… the same plurality of individuals may simultaneously satisfy two or more sets of group-constituting conditions. The groups may therefore coincide in membership without being identical – extensionality goes.” -- 11 of 108 -- The Synthesis of a TOVE Persons Ontology Page 10 Copyright © 2003 - Chris Partridge. All rights reserved. literally10 , they suggest that organisations are of different inconsistent types. How can a ‘set of constraints on activities performed by a set of collaborating agents’ be something that consists of three things; the same ‘set of Organisation-Agents (said to be members of the organisation)’, a set Organisation-Units and an Organisation-Goal tree? An added problem in this case is that the descriptions do not tie in with our raw intuitions, which do not regard individual organisations, such as IBM, as sets of any kind. 4.1.2 People’s intuitive notion of organisation The analysis needs to start with a reasonably clear idea of what an organisation is. TOVE does not really supply this, though it gives us some raw material. The STPO makes an initial attempt to develop one now, by fitting organisation into a taxonomy. 4.1.2.1 Organisations are elements It seems plausible to say that individual organisations, such as IBM, are not types with instances (or properties with instances or sets with members or universals with exemplifications, or whatever mechanism there is in the top ontology to deal with this kind of fundamental particular-universal distinction). This seems relatively innocuous. How could IBM as an organisation be a type that had various IBMs as its instances? And it ties in with the way people seem to regard organisations. This means in terms of the STPO typonomy that organisations are elements. 4.1.2.2 Organisations are concrete Element is a very general type. The STPO can be more specific and say that organisations are also concrete elements: in the sense that they have temporal and spatial extents. This is not an original view, (Smith 1999), for instance, says “Agglomerations [which includes organisations] are, ontologically speaking, spatial objects. Their lives or histories are spatio-temporal objects. … Agglomerations are spatial objects which inherit their spatial properties from the spatial properties of the relevant members or participants.” Adopting this view is important for the STPO analysis, as it provides the basis for investigating organisations’ spatial and temporal boundaries. In some contexts, we ignore an organisation’s physical extent in way we do not with prototypical physical objects. It is not as easy to kick (or pick up and throw) a company as it is a stone. However a little reflection reveals their concrete characteristics. The temporal extent of modern companies is particularly clear. Companies are clearly founded. For example, IBM was founded as Computing- Tabulating-Recording Company (C-T-R) in 1911 – as a result of the merger of the International Time Recording Company, Computing Scale Company, and the 10 We are not claiming that the builders of TOVE intended their informal descriptions be taken so literally. We are looking for a literal explanation and the problem is that TOVE does not give one. -- 12 of 108 -- The Synthesis of a TOVE Persons Ontology Page 11 Copyright © 2003 - Chris Partridge. All rights reserved. Tabulating Machine Company – and renamed International Business Machines Corporation in 1924. Companies also clearly go out of existence – as the International Time Recording Company, Computing Scale Company, and the Tabulating Machine Company did in the merger that founded IBM. If we were to look at the founding of IBM in more detail, we would find the activities that constitute the founding – meetings, document signings and registrations and so on. Similarly organisations have a spatial extent. People talk about companies being in a place. For example, IBM’s website says it was based in New York City when it was founded. Though when a company is particularly scattered – as IBM is today – it is not informative to state its complete spatial location. For example, it would seem strange to say that IBM is on the planet Earth – though this is not false, just not useful. But people still talk about parts of a scattered company – for example, saying that IBM has a presence in South Africa. Some people may be feeling concerns about the vagueness of the spatio-temporal boundaries of organisations, particularly at the granularity of everyday medium- sized objects, such as ourselves. However, vagueness is a general problem that dogs most, if not all medium sized objects – it is not specific to organisations. For example, the boundaries of medium size objects, such as cats, stones and chairs are not clear at the microscopic level. Vagueness only really becomes of practical interest if it hinders us from doing something. Nevertheless, in the case of organisations, there is useful work that can be done to crispen their boundaries. For example, it is clear that organisations’ spatio- temporal extent is, in many cases, related to, maybe dependent on, the spatial (and sometimes temporal) extent of the people that are participating in (part of) them. Similarly the temporal extent seems dependent upon the temporal position of events – such as a companies founding or its dissolution. It is much less clear what the nature of this dependency relation is. One of the aims of the STPO analysis is to clarify this. 4.1.2.3 Organisations are physical The STPO also assumes, at least as a working hypothesis, that organisations are physical: in the sense that they are material, composed of matter. (Searle 1995) talks of institutional facts [which include organisations] needing to have a physical realisation. Using money as his example, he says (p.34) “[J]ust about any sort of substance can be money, but money has to exist in some physical form or other. … What is true of money is true of chess games, elections, and universities. All these can take different forms, but for each there must be some physical realisation.”11 The STPO finds this position congenial as it puts organisations and 11 This passage does not make clear whether John Searle is claiming that the representation or the represented needs to be physical. The position that the representation is physical is reasonable. The position that the represented is physical needs some explanation, particularly in the case of synthetic assets such as financial options. This will be part of the task of the Asset stage of the synthesis. -- 13 of 108 -- The Synthesis of a TOVE Persons Ontology Page 12 Copyright © 2003 - Chris Partridge. All rights reserved. human beings (both persons) in the same category and explains simply how organisation have causal properties. It also sharpens the questions that can be asked about identity conditions. Within the CEO’s meta-ontology there is no distinction between element, concrete and physical – these collapse into a single type. However, for now in the STPO analysis we include these as separate elements, when the synthesis analysis is relatively complete it should be clear that nothing is lost by collapsing these into single type – and that the result is a simpler structure. 4.1.2.4 Organisations are persons From an enterprise perspective, one of the most important distinguishing features of an organisation is that it is a person. What unifies the categories of organisation and natural person is that their instances have the power to enter into transactions on their own behalf, intentionally acquiring rights and obligations. This is central to the CEO’s notion of persons. What distinguishes them is that organisations instances are intentionally constructed, whereas natural person’s, we assume, are not 12 . 4.1.2.5 Persons are physical This raises the taxonomic questions as to whether persons – things with the capability to acquire rights and obligations – are also physical. Persons includes humans, what might be called natural persons, as well as organisations. Natural persons are clearly physical. Theologians may want to argue that God is a person and neither physical nor concrete. However, God rarely plays an explicit part in the context of enterprise systems, so it would seem feasible to regard all persons as physical and concrete. As a working hypothesis the STPO assumes this. 4.1.2.6 Re-identification Acquiring rights and obligations implies an important practical characteristic for persons and hence for organisations: that it is important to be able to re-identify them over time. Persons enter into transactions with an organisation and so acquire rights or obligations with respect to it. They would, understandably, be uncomfortable if they could not re-identify it. Furthermore, the members of an organisation need to be able to re-identify the organisation of which they are members. For this to work, it must not just be possible in principle, but recognised by people as reasonably easy to do in practice. Among other things, the intentions that underwrite an organisation need to inspire a degree of confidence that this is likely to happen. 12 Ignoring for the moment the arguments that suggest that natural persons (unlike their natural bodies) are also intentionally constructed. For a summary of these see Olson (1994) The world on paper pp. 238-9 – and for a more detailed argument see Snell (1982) The discovery of the mind. -- 14 of 108 -- The Synthesis of a TOVE Persons Ontology Page 13 Copyright © 2003 - Chris Partridge. All rights reserved. 4.1.2.7 Organisations persist through time The need for re-identification re-affirms the intuition that an entity that is an organisation has to persist in time. That ‘being an organisation’ cannot be explained as an attributive property like ‘being red’ or ‘being tall’, which can be true at a moment in time, irrespective of the past and future. It also links into another common feature of organisations, that they can come into and go out of existence – that they are born and that they usually die13 . The original intentions bring the organisation into existence. Typically, more formal organisations are constructed in a framework that spells out in detail how this commitment should be made: legal company creation is an example. In less formal cases, there may not be an easily identified agreement event. Certainly there are no rules enforcing an explicit agreement event. What sometimes seems to happen is just a growing awareness among the people involved that they are somehow agreeing to act together – as a person. 4.1.2.8 The emerging STPO taxonomy The STPO taxonomy is emerging. Organisations are concrete, physical elements. They are also Persons. This raises the question of whether Person can also be classified as concrete, physical elements. It seems a reasonable hypothesis that Persons are concrete elements. However, given that Persons are causal agents, the STPO makes the stronger assumption that Persons like Organisations are also physical elements. The final STPO ontology does not depend upon this assumption, but making it helps to focus its analysis. 4.1.3 Organisations are intentionally constructed Organisations are not the prototypical example of concrete physical things, so it helps to characterise what makes them different. A common explanation is that they are (types of) intentionally constructed objects. 4.1.3.1 Explanation in terms of intentions (Searle 1983; 1995) offers an explanation of organisations in terms of collective (social) intentionality. (Searle 1983), on p.4, suggests that what marks intentionality is a representation that is directed towards an object. (Searle 1995), on p.43, suggests that what marks institutional objects is that they obey a rule of the form “X counts as Y for C” – for example, that this piece of paper counts as a £10 note for the English. Here the collective intentionality of the English underwrites the existence of the £10 note. This involves, at least, belief (which involves representation). For example, if the English stopped believing that the English currency was money – it would stop being (in some sense) money. But more than belief is necessary, there also needs to be a commitment to accepting 13 Organisations can, in principle, be immortal, but they usually are not. -- 15 of 108 -- The Synthesis of a TOVE Persons Ontology Page 14 Copyright © 2003 - Chris Partridge. All rights reserved. the money as payment for goods – and so on. This principle is, in fact, enshrined in English Law – where shopkeepers are obliged to accept legal tender. (Searle 1995), on p.118, also recognises that the creation of many institutional objects is marked by the performance of a ritual – for example, marriage is marked by a marriage ceremony. Typically it is specific intentional events, focussed on the intentional objects that mark the temporal boundaries of institutional objects. For example, a marriage can be formally dissolved by a divorce (an intentional legal ritual). However it can also be ended by a physical catastrophe that kills one or both of the partners14 . Identifying and classifying the events that mark the temporal boundaries of organisations will provide us with the basis for their persistence conditions. (Bratman 1987; 1999a) also offers the material for an explanation in terms of intentions – where he uses this term in its everyday sense of ‘a determination to act in a certain way’. Michael Bratman focuses on the use of plans in practical reasoning and describes the nature of plans in terms of partial intentions. People make plans because, in practice, human reason is too limited to be able to examine every situation anew as it occurs – so they (we) have to spread the load by planning in advance. Obviously, these plans cannot cover every detail, as some aspects will be left unspecified. Using his example: Anne may plan to go to the cinema tonight without specifying which cinema or what time, because she does not yet know when she will finish work. When we adopt a plan we are making an intention to carry it out. This intention is a representation in John Searle’s sense (it is directed at something) and, as Michael Bratman is at pains to point out, it has more commitment than a mere belief or description. Again using his example: Arthur can believe that he is going to be shot at dawn tomorrow without making any commitment to make sure this happens. However, if Arthur intends to be shot at dawn, we expect him to have some sort of commitment towards making this happen. Plans are also useful when trying to co-ordinate the activities of a group – where it is impractical to specify everything in advance, particularly if circumstances can change. They are a mechanism for generating a common commitment without having to spell out the details in full. This has clear parallels with Margaret Gilbert’s notion of a social group’s joint readiness to do something15 . She also clearly explains how this readiness is selective, picking only relevant activities. How even if (using her example) the Library Committee and the Food Committee have the same members, when they meet as the Library Committee they discuss the purchase of books not menus. This feature makes it difficult to accept that organisations or social groups are simply constituted by their members. 14 These kinds of physical catastrophe also destroy parts of the intentional agreement. In the marriage example, the destroyed person’s mental representation of the marriage – and so have an intentional aspect. 15 Explained on pp.185-6 of Gilbert (1992) On social facts. -- 16 of 108 -- The Synthesis of a TOVE Persons Ontology Page 15 Copyright © 2003 - Chris Partridge. All rights reserved. 4.1.3.2 Irreducibility of collectivities Gilbert takes a stronger view – not only are social groups not simply constituted by their members, but that they are not reducible to merely the actions of their members acting independently. The actions need to be considered as, in some sense, joint. Searle takes a parallel view with regard to collective intentionality – that it also is not reducible to singular intentionality. While the irreducibility thesis seems correct, and Gilbert’s arguments are persuasive, some elements of the context that both her and Searle give need extending for an enterprise ontology. 4.1.3.3 Non-collective (singular) organisations Gilbert focuses on plural subjects. As this stands, it is not a wide enough basis for an analysis of what an organisation is. This is because, in practice, there are many organisations that are not plural subjects, not collectivities. Classic examples are sole proprietorships and corporations sole. The final ontological model of the enterprise has to recognise that organisation includes these singular (non- collective) subjects. Sole proprietorships are the oldest, most common and simplest form of business organisation: they are business entities owned and managed by a single person. A corporation sole consists of one incorporated office and provides for a succession of office holders. This entity has been recognised by English law for over 500 years. This form of corporation grew out of a need to institutionalise the monarchy, separating what the person occupying the throne did (acting ex officio) from what the monarch did (acting in officio) – and to help ensure that the commitments of the monarch did not die with the person occupying the throne. The legal maxim “The monarch never dies” is true, because “the monarch” is an office, not a human. The same form was subsequently used by the Church to find an orderly and secure way to hold and pass title to church property. The English Crown and the Archbishop of Canterbury are both examples of corporations sole. As we shall see later, other types of offices (positions), such as Chairman and Managing Director, are also kinds of organisation – with similar notion of acting in officio and ex officio. What seems to characterise these single, often serial, subjects, like their plural counterpart, is that they are intentionally constructed. Some kind of intentional agreement provides the central unifying principle that pulls together their activities. The issue is whether Gilbert’s analysis can be extended to include these cases. In Gilbert’s case this means extending it to cover non-collective (singular) subjects. One way to do this is to look at borderline cases of plural subjects that come close to being singular. One such case is where the subject is not always plural. One could find situations where a team of soldiers undertake a mission and all but one are killed before the final soldier completes the mission. It seems to make more sense to regard the single person carrying out the joint commitment as still being a part of the team – though the only part at that time – than regard the -- 17 of 108 -- The Synthesis of a TOVE Persons Ontology Page 16 Copyright © 2003 - Chris Partridge. All rights reserved. team and its mission as disappearing as the second to last member of the team dies. One can imagine a less convincing case, where a subject is never plural, though it might have been. Consider, a case where someone starts up a team with a clear objective: for example an expedition to the South Pole. If the team is assembled and reaches its objectives, one would regard the early efforts of the sole member as part of the team’s preparation. However, what happens if a team is not assembled and the ‘sole member’ decided to undertake the expedition by him or herself. In both cases we talk about an expedition. In the second case, the potential for plurality seems to give some weight towards regarding it as a valid subject. It seems to me that this analysis tends to show that plurality and singularity are not a essential features of subjects. There are many cases where subjects are at one time singular and another plural. There are also many cases where they have always actually been plural – but it is possible that they might have been singular. This leads me to believe that one can apply much of the Gilbertian analysis to the singular case. 4.1.3.4 Non-collective (singular) intentionality Searle focuses on collective intentionality – where the subject of the intentionality is a collectivity, a plural subject. The Searlian analysis divorces the collectivity underlying collective intentionality from the (socially) intentionally constructed object. This places no barriers to the existence of singular subjects. One example he offers is ‘Miss Alameda County’16 . For Searle, these singular subjects are constructed (and maintained) by collective intentionality. So it does not exclude the classic enterprise cases of singular subjects. However, there is one aspect of the analysis that needs clarification. Does it imply that the intentionality that constructs singular subjects necessarily has to be collective? If so, this would exclude a Robinson Crusoe from constructing singular subjects. However, there nothing in Searle’s analysis that specifically implies that this is so – it is more that his focus is on collective intentionality. (Searle recognises that there are ‘singular intentional facts’, but he states that his interest is in the institutional objects that involve collective intentionality17 and so singular intentionality is not much discussed.) With our focus on the enterprise, rather than collective intentionality, we need to include a wider range of intentionality, one that encompasses both singular and collective varieties. It seems that for organisations, at least, the key element is what the intentions are, rather that whether they are collective. 16 See p.96 of Searle (1995) The construction of social reality. 17 Ibid. The construction of social reality, on pp.121-2, “distinguishes between singular intentional facts, such as I want a drink of water now, and collective intentional facts, such as the fact those hyenas are attacking a lion.” He goes on to identify institutional facts with collective intentional facts and social facts as a special subclass of these. -- 18 of 108 -- The Synthesis of a TOVE Persons Ontology Page 17 Copyright © 2003 - Chris Partridge. All rights reserved. 4.1.3.5 Plural subject’s members’ intentions (Gilbert 1992), on p.199, proposes what might be called the ‘*we test’ for the existence of a plural subject18 : “A person X’s full-blooded use of ‘we’ in ‘Shall we do A?’ with respect to Y, Z and himself, is appropriate if and only if it expresses his recognition of the fact that he and the others are jointly ready to share in doing A in relevant circumstances.” She also says that “it is assumed that people are jointly ready if and only if it is common knowledge among them that each has openly* manifested to all his quasi-readiness to share some action in the circumstances in question.” Gilbert (as does (Bratman 1999a)) builds upon an idea of (Lewis 1969) that some kind of common knowledge of the intention, or at least its main elements, is an aspect of the joint readiness (a kind of collective intentionality). This makes sense if one wants to link responsibility to knowledge. While this is a good characterisation of many social groups, particularly of the sort Gilbert uses for illustrative purposes – it is less suitable as a characterisation of enterprise organisations. The trouble is that it assumes that every participant in a plural subject has the common knowledge and is jointly ready – that every participant has to have the right level of intentionality. Gilbert realises this and explains (on pp.230-2 of (Gilbert 1992)) why a firm that may not be, in her terms, a plural subject either because it has excluded common knowledge by largely automating its operations (not enough common knowledge) or because it is a non- harmonious collection of people (not enough commitment to joint readiness 19 ). She suggests that the difference between her characterisation and the enterprise’s may lie in the distinction between familial and business relations20 . It is clear that there is no requirement on all participants in enterprise organisations to commit to joint readiness or to have the common knowledge needed to support it. The work on information economics suggests that in large organisations this kind of common knowledge would be the exception rather than the rule: see, for example, Kenneth Arrow’s work. Organisations do not have to be harmonious or unautomated – cynics may note that a lack of harmony and high levels of automation are common in many modern organisations. Clearly all organisations are not plural subjects in Gilbert’s strict sense. 18 Gilbert regards the use of ‘we’ in ordinary discourse as a good test for plural subjects. However she suggests the use of ‘*we’ as there is no guarantee that ‘we’ will only refer to plural subjects. Note that the ordinary language ‘*we test’ works for organisations generally, including non- collective organisations. For example, the English Monarch traditionally uses ‘we’ when speaking in officio. 19 Nevertheless, as observed in the previous footnote, these organisation pass the (admittedly rough) Gilbertian ‘*we test’. Employees typically use the ‘we’ pronoun when speaking on behalf of the company – despite automation and disharmony. 20 For which sociologists have offered explanations: Tönnies’ distinction between community and society and Weber’s between communal and associative relations – see e.g. p.230 of Gilbert (1992) On social facts. -- 19 of 108 -- The Synthesis of a TOVE Persons Ontology Page 18 Copyright © 2003 - Chris Partridge. All rights reserved. Gilbert discusses (on p. 232-3) a phenomena she calls ‘complex groups’, in which she accounts for some of the features of enterprise organisations. She notes that someone can join a club without being aware of its purpose, and commonly be regarded a member of the club. Gilbert argues that the joining of the club indicates, presumably, “a willingness to jointly pursue with the other members whatever as yet undisclosed purposes the club has.” She also notes strategies for delegating the responsibility for the details of decision making. However, I suspect that the difference between these kinds of enterprise organisation and Gilbertian plural subjects has been exaggerated. This becomes clearer if one examines another case of complex groups noted by Gilbert: infants. It seems right to say that these are not yet capable of the kind of common knowledge needed for a Gilbertian plural subject but that this does not seem to hinder them from being parts of families, which we normally regard as good examples of social groups21 . Gilbert argues that she is “isolate[ing] the phenomenon which is central to collectivity existence as intuitively understood.” It seems to me a more common sense position is to regard infants’ membership of the family as the simple central case and common knowledge as a more peripheral sophisticated case – though one that may now be relatively common in educated adult society. The argument from common knowledge assumes that there must be some form of representation of the membership in the minds of the members. Consideration of animals and humans’ unreflecting behaviour suggests that the tendency to act together is instinctive – maybe, in some senses, more natural than acting alone22 . Searle makes a similar point, proposing a non-representational background of attitudes and predispositions underlying most forms of social construction23 . In many cases, the original ties that bind together collectivities may be instinctual and only later become rationalised into notions of common knowledge and joint readiness. If this is so, then non-harmonious organisations and ignorant members (including infants) are unproblematic. However, explaining and accounting for responsibility may become more problematic – as it is in real life. 21 Though, as Ibid. p.234 points out, not everything one might wish to call a family would qualify as a social group. And the membership of the group is not always clear; one could question whether a ‘black sheep’ who has had his or her connections with the family severed was a member. 22 As Hume (1739-40) A treatise of human nature notes “Everyone has observed how much more dogs are animated when they hunt in a pack, than when they pursue their game apart. We might, perhaps, be at a loss to explain this phenomenon, if we had not experience of a similar in ourselves.” Book II Of the Passions, PART II Of love and hatred, SECTION XII - Of the love and hatred of animals. 23 And this non-representation aspect of socially constructed objects needs to be recognised analysing them – see the discussion of this in Partridge (2002b) A new foundation for accounting – which also notes the analogies with discussions of rule-following in recent philosophy. -- 20 of 108 -- The Synthesis of a TOVE Persons Ontology Page 19 Copyright © 2003 - Chris Partridge. All rights reserved. 4.1.3.6 Adding intentionally constructed objects to the STPO taxonomy The purpose of the preceding sections is to make a case for including the insights of Bratman, Gilbert and Searle into the explanation of what an enterprise organisation is. It is also making a case for regarding the full range of groups and organisations as a unified category. It is not intended to provide final arguments – there is clearly much more work to do here. However, it provides a good working hypothesis for the STPO. The main elements of the preceding analysis need to be incorporated into the STPO’s taxonomy. The type, intentional object (AKA intentionally constructed object), needs to be recognised as does the fact it is a sub-type_of organisation. This raises the question of how it relates Intentional Object to the rest of the taxonomy. A quick analysis reveals no clear answer. It may seem a modest assumption that the instances of Intentional Objects are all Elements – in other words, that Intentional Object is a sub-type_of Element. But there are cases, such as Sterling, that are clearly both types and Intentional Objects. And if Intentional Object has a type instance, Sterling, which is not obviously physical or concrete, then it seems that it cannot be a sub-type_of Physical or Concrete. For now, as it seems to be irrelevant to the Persons analysis, the STPO leaves Intentional Objects as a sub-type_of Object. The finer details of this level of the taxonomy can be decided later – when there is more evidence. The notion of organisation has a number of related senses. For the sake of simplicity and regimentation, and because it serves the CEO’s purpose, the STPO proposes to identify organisations with intentionally constructed persons. From a taxonomy perspective, this make organisation the intersection of Intentional Object and Person. As the review of the STPO analysis proceeds, it should become clearer why this is a sensible decision. 4.1.4 Precisifying organisation At a simplistic level, people lump organisations, social groups and other collectivities together with a host of other types of things in a category labelled collections. They also tend to identify the collections, including organisations, with their members. From this perspective, it may feel as though shifting the burden of maintaining identity from the members to an intentional agreement is merely postponing the hard analysis, or even worse, a case of “‘postulating’ what we want”, which as Russell24 memorably said has “the advantages of theft over honest toil.” However, the need to precisify organisation makes the simplistic notion of collection that identifies it with its members an untenable basis for the identity of organisations. 24 On p.71 of Russell (1970) Introduction to mathematical philosophy. -- 21 of 108 -- The Synthesis of a TOVE Persons Ontology Page 20 Copyright © 2003 - Chris Partridge. All rights reserved. 4.1.4.1 Organisation as a collection of members As a number of commentators have pointed out, naïve set theory textbooks provide good examples of the simplistic notion of collection, where collection is regarded as a synonym for set. For example, on p.177 of (Suppes 1957) we find: “By a set we mean any kind of collection of entities of any sort … Many other words are used synonymously with ‘set’: for instance, ‘class’, ‘collection’, and ‘aggregate’.” Page one of (Halmos 1960) offers “a pack of wolves, a bunch of grapes, or a flock of pigeons are all examples of sets of things … To avoid terminological monotony, we shall sometimes say collection instead of set. The word ‘class’ is sometimes used in this connection.” (See pp.91-2 of (Marcus 1993) for more examples.) From this simplistic perspective, organisations are regarded as a collection of their members. This is perhaps not surprising when a common definition of collection is ‘a group of things that have been gathered together’. For the more sophisticated applications that the CEO is interested in, more precision is needed (relevant). As Ruth Marcus notes the simplistic approach “obscures commonplace distinctions that we make between assortments, collections, and classes.” It ignores the distinction between mereological sums and sets25 , what used to be called, pre-Frege, distributive and collective sets (TOVE’s informal descriptions do the same). It does not recognise that the identity conditions for organisations have to be able to differentiate between two organisations with the same members. It provides no explanation for the way in which the membership of a ‘collection’ (whether mereological sums, sets or something else) shifts over time – and how this affects the identity conditions26 or how to differentiate between organisations with one member and the member. It also cannot explain organisations with no members. Finally, it obscures the distinction people make between an organisation or social group and the collection of its members at a time. Where, for example, it makes sense to say: these two organisations (or social groups) have exactly the same (collection of) members – implying that there are three things, not one or two. 4.1.4.2 Organisation’s identity The precisification of the notion of organisation, in terms of intentional agreement, developed in the STPO explains these distinctions. What unifies the organisation is its on-going intentional agreement. Someone becomes a member of the organisation, when they become a (member-)party to 25 Margaret Gilbert also seems to be uninterested in the distinction – she says (on p.430 of Gilbert (1992) On social facts “One may be tempted to think of societies as mereological sums of singular agents, or as sets of singular agents, in the logician's sense. Though strictly speaking it may be correct ...” 26 For more on this problem see Ch. 7, §3 The shift to object semantics in Partridge (1996) Business Objects: Re - Engineering for re - use. -- 22 of 108 -- The Synthesis of a TOVE Persons Ontology Page 21 Copyright © 2003 - Chris Partridge. All rights reserved. the agreement. From then on, a member’s activities are part of the organisation’s activities if they are included as such under the intentional agreement. An interesting limit case that illustrates this point is where someone, Jane say, runs two sole proprietorships. In this case, she is party to two different intentional agreements. And these, typically, are associated with different activities: in other words, when Jane is checking the books for the first sole proprietorship, she is not doing anything for the second sole proprietorship. Note that both organisations may include book-checking activities. It is not the type of the activity that ultimately determines which organisation it is part of – but whom the activity is for. The intentional agreement is what gives the organisation its identity. The intentional agreement (not the members per se) is what preserves the organisation’s identity through time. Members can join and leave an organisation without affecting its identity – but terminate the intentional agreement and the organisation ceases to exist. The notion is fruitful in explaining seemingly difficult cases, such as the difference between two companies that have the same members. There are two different intentional agreements. The intentional agreement is also important for determining the spatio-temporal extent. The agreement often directly gives us a good basis for determining the organisation’s temporal extent. The beginning is, and the end often is, an intentional performance (Searle’s term) focused on the agreement – such as legal company registration. In abnormal cases, the end may be the result of an unrelated catastrophe – which destroys the intentional agreement. As, for example, in many legal jurisdictions, the death of a partner dissolves the partnership. Furthermore, intentional agreement indirectly determines the spatial extent – by licensing selected members’ activities, and these, as activities of the organisation, help to determine its spatial extent. 4.1.4.3 Intentional agreement’s identity and extent The analysis so far has shown that the intentional agreement is a far more plausible candidate as a basis for an identity condition than the members. However, to completely escape a (Russellian) accusation of “‘postulating’ what we want”, there needs to be sufficient confidence that an adequate characterisation of intentional agreements can be made – as it seems that there is a clear understanding of what a member is27 . The organisation is not identical with, but depends upon the intentional agreement. From the perspective of Persons, we do not need a taxonomy of intentional agreements, but some idea of what an adequate characterisation would be. Intentional agreements are a species of Transaction – another segment of the CEO synthesis. The transaction segment will develop the taxonomy of intentional agreements. 27 The STPO analysis of member, described later, shows that the understanding is not quite as clear as it seems. -- 23 of 108 -- The Synthesis of a TOVE Persons Ontology Page 22 Copyright © 2003 - Chris Partridge. All rights reserved. A good starting