A Survey of Top-level Ontologies - Appendix J

Ontological commitments – technical details 

In this appendix, we cover a range of the relevant background detailed technical issues. We also include relevant extracts at the end, as a gateway to the wider literature on this topic. 

J.1 Natural language ontology – foundational ontology 

One use of ontology is to describe the language (and associated concepts) we use to talk about things – hence a natural language ontology. Friederike Moltmann gives a good outline of this in the extract below; noting, for example, that it “is better to be characterized as the ontology that a speaker implicitly accepts when using the language” and so “may include merely conceived objects besides objects that happen to be actual ones.” 

She also contrasts this with “the ontology of what there really is” which she, following Fine (2017), calls ‘foundational ontology’. There, Fine contrasts foundational ontology with what he calls naïve ontology. As Moltmann points out (see extract below I.1.1), “Fine’s (2017) term ‘naïve metaphysics’ thus is misleading” – natural language or descriptive ontology is a better description. Fine (2017) also suggests that “naive metaphysics will be the metaphysics of appearance while foundational metaphysics will be the metaphysics of reality, with the one concerned to discern the nature of the world as it presents itself to us and the other concerned to discern the nature of the world as it is in itself.” 

Note, in earlier ontological commitment analyses (Borgo (2002), Semy (2004) – extracts below – and Orbst (2010)) there is a similar distinction between descriptive and revisionary ontologies (based upon Strawson’s (1959) distinction). This nomenclature then evolved to descriptive and realist ontologies in ROMULUS (2013). However, as you can see from the extracts, this distinction was often expanded to cover with linked distinctions such as between space and time and space-time, and endurants and perdurants. 

J.1.1 Friederike Moltmann – Natural Language Ontology 

Extract from: Friederike Moltmann (2020) Chapter 13 – Abstract Objects and the Core-Periphery Distinction in the Ontological and the Conceptual Domain of Natural Language. Online: https://www.academia.edu/38640396/

13.3 Natural Language Ontology 

The following is a brief outline of the discipline whose subject matter is the ontology of natural language, natural language ontology. With its referential noun phrases, which take entities as semantic values as well as with its lexical predicates and constructions that involve entities in other ways, natural language reflects an ontology. This ontology is the subject matter of a particular branch of metaphysics, that of natural language ontology. More specifically, natural language ontology is part of descriptive metaphysics in Strawson’s (1959) sense or what Fine (2017) calls ‘naïve metaphysics’. Descriptive metaphysics has as its subject matter the ontology reflected in our ordinary judgments. Natural language ontology has as its subject matter the ontology reflected in linguistic intuitions, that is, judgments about the acceptability or grammaticality of natural language sentences and constructions. 

What is important about descriptive metaphysics is that it is not about the ontology of what there really is. This is instead the subject matter of a different branch of metaphysics, what Fine calls ‘foundational metaphysics’.[4] Descriptive metaphysics and natural language ontology in particular concerns itself with how things appear, given the data, without addressing the question of whether they are real (which is to be addressed only by foundational metaphysics). For natural language ontology; this means that no foundationalist consideration should come into play when positing objects as semantic values, such as assumptions as to whether those objects really exist (in the sense of being fundamental) or what they may be reduced to. More important is what sorts properties the semantic values of referential noun phrases may have, as is reflected (at least to an extent) in the applicability of types of natural language predicates. The domain of objects in the ontology of natural language thus may include merely conceived objects besides objects that happen to be actual ones. (This will also be important for how to make sense of the expandability of the ontological domain of the language through technical or philosophical discourse in Sect. 13.4.) 

The subject matter of natural language ontology is not the ontology that ordinary speakers (non-philosophers) naively accept when thinking about what there is. [5] The latter is the subject matter of folk metaphysics, not natural language ontology. What speakers accept when they reflect does not matter for natural language ontology. Natural language ontology rather deals with the ontological categories, notions, and structures that are implicit in language whether or not speakers would accept them upon reflection. The ontology of natural language thus is better to be characterized as the ontology that a speaker implicitly accepts when using the language and as such is distinguished from both the reflective ontology of ordinary speakers as well as philosophers and the ontology of what there really is (Moltmann 2017, 2019). 

[4] For Strawson (1959), descriptive metaphysics rather contrasts with what he calls ‘revisionary metaphysics’. The aim of revisionary metaphysics, for Strawson, is to conceive of a better ontology than the one we ordinarily accept. (Strawson does not specify further how ‘better’ is supposed to be understood.) 

[5] Fine’s (2017) term ‘naïve metaphysics’ thus is misleading, and ‘descriptive metaphysics’ a better term to use for the branch of metaphysics that comprises natural language ontology. 

J.1.2 Wonderweb deliverable d15: Ontology Roadmap (2002) 

2 Design Options and Ontological Choices 

Before addressing specific issues about domain of discourse, basic categories, and their relations [4], it may be important to clarify the general attitude towards ontological analysis, or – in other words – the motivations and the constraints that drive our conceptualization of reality. It comes to no surprise that the design options for building foundational ontologies reflect the main categorical distinctions discussed in philosophy. However, among all the philosophical stances and distinctions, foundational ontologists seem particularly interested in two general attitudes: a) descriptive vs. revisionary, and b) multiplicative vs. reductionist. 

(a) A descriptive ontology aims at capturing the ontological stances that shape natural language and human cognition. It is based on the assumption that the surface structure of natural language and the so-called commonsense have ontological relevance. As a consequence, the categories refer to cognitive artifacts more or less depending on human perception, cultural imprints and social conventions. Under this approach, there are no major restrictions on the postulation of ontological categories because overall philosophical or scientific paradigms are neglected. This attitude stands in contrast to the revisionary approach. The revisionist considers linguistic and cognitive issues at the level of secondary sources (if considered at all), and does not hesitate to paraphrase linguistic expressions (or to re-interpret cognitive phenomena) when their ontological assumptions are not defensible on scientific grounds. The following example should make this contraposition clear. Commonsense distinguishes between things (spatial objects like houses and computers) and events (temporal objects like bank transfers and computer repairs). In the wake of relativity theory, how- ever, time is viewed as another dimension of objects on a par with the traditional spatial dimensions. Considering the consequences of this scientific theory (or theories), some philosophers and computer scientists have come to believe that the commonsense distinction between things that are and things that happen should be abandoned in favor of a unified viewpoint. According to these revisionist researchers, everything extends in space and time, and the distinction between things and events is an (ontologically irrelevant) historical and cognitive accident. This example shows that a revisionary ontology is committed to capture the intrinsic nature of the world by providing structures that are independent from the conceptualizing agents. 

[4] A. Gangemi, N. Guarino, C. Masolo, and A. Oltramari. Understanding top-level ontological distinctions. In IJCAI-01 Workshop on Ontologies and Information Sharing, pages 26-33, Seattle, USA, 2001. AAAI Press. http://SunSITE.Informatik.RWTH-Aachen.DE/Publications/CEUR-WS/Vol-47/. 

J.1.3 MITRE: Toward the Use of an Upper Ontology for U.S. Government and U.S. Military Domains (2004) 

3.2.1 Descriptive vs. Revisionary 

Descriptive and revisionary ontologies [67], [49] are based on ontological stances or attitudes towards the effort of modeling ontologies, i.e., how one conceptualizes the world and what an ontological engineering product is or should be. A descriptive ontology tries to capture the more commonsensical and social notions based on natural language usage and human cognition, emphasizing the agent who conceives and deemphasizing scientific and philosophical considerations. A revisionary (sometimes called prescriptive) ontology, on the other hand, does emphasize (or even, strictly adheres to) the scientific and philosophical perspectives, choosing to base its constructs and modeling decisions on scientific theories and a philosophical stance that tries to capture the world as it really is (it prescribes the world), and not necessarily as a given historical agent conceives it to be. A revisionary ontology therefore says that its modeling constructs are about real things in the world as it is. 

In practical terms, all of the constructs in a revisionary ontology will be space-time objects, i.e., necessarily having temporal properties; in a descriptive ontology, that will not be the case. In the latter, entities (sometimes called endurants, but perhaps better called continuants) such as “hammer” and “tank” that have only incidental temporal properties and events (processes, actions, activities, etc., sometimes called perdurants, but perhaps better called occurrents) such as “attacking” and “cashing a check” that have explicit temporal properties, are modeled with or without those temporal properties, respectively. Often in natural language there are two correlated forms/usages that express the distinction: the nominal and the verbal. A nominal (noun) “attack” is expressed as in “The attack on the enemy began at 600 hours.” A verbal (verb) “attacked” is expressed as in “We attacked the enemy at 600 hours.” 

[67] Strawson, P.F. 1959. Individuals: An Essay in Descriptive Metaphysics. London: Methuan University Press. 

[49] Obrst, L., H. Liu, R. Wray. 2003. Ontologies for Corporate Web Applications. Artificial Intelligence Magazine, special issue on Ontologies, American Association for Artificial Intelligence, Chris Welty, ed., Fall, 2003, pp. 49-62. 

J.2 Extensional and intensional criteria of identity 

J.2.1 Making a single broad choice 

Typically, top-level ontologies will adopt one of these broad choices for their main entities, though they may adopt the alternative for selected entities. 

Typically, ontologies with an extensional strategy pick very general groups as the bearers of criteria of identity. BORO, ISO 15926, HQDM and IDEAS are examples of this. They include types/sets with an extensional membership criterion and particulars with a spatio-temporal parts criterion (often simplified to spatio-temporal extension). 

Typically, ontologies with an intensional strategy pick less general, more specific groups, often known as sortals, as the bearers of criteria of identity, as the essential characteristics that inform the object’s identity are less general. BFO, DOLCE and UFO are examples of this. 

J.2.2 Bearers of specific identity criteria 

Often entities are labelled with their identity choice; so extensional entities and intensional entities. As the examples above show, for various reasons (including pragmatism) these choices are not made for each individual entity, rather they are made for groups of entities, where the choice applies to all members of the group. Often the entities are grouped by their specific identity criteria, where different groups have different criteria – the groups bear these identity conditions. Normally these groups are disjoint, as otherwise there is a requirement for meta-rules to arbitrate for objects that fall under two different criteria of identity (Guarino (2000) offers this meta-rule: Properties carrying incompatible Identity Criteria are necessarily disjoint). 

J.2.3 Taming intensional promiscuity 

We can illustrate the core difference between the two ways of capturing identity with an example – showing the potential promiscuity of the intensional approach. As already noted, general objects have an extension; the objects that instantiate the object. Under extensional criteria, this extension determines its identity. Under intensional criteria, it does not. Consider the earlier example of an equilateral triangle – defined as having equal sides. Now consider equiangular triangles – with three equal angles. One can prove that these definitions are equivalent. However, they are different, so it makes sense to say they are different intensions, even though one can logically show they always have the same extension. One can easily develop more and more baroque definitions that have this same extension. Under a permissive intensional strategy, these would all be different objects. One is then faced with deciding what counts as a different meaning. As far as we have been able to determine, in so far as the candidate top-level ontologies have adopted an intensional strategy, they have adopted a permissive one. 

J.2.4 Extensionality as varieties of formal grounding 

One can view extensional identity as a form of grounding. In the case of a set, the set is composed of and grounded and dependent upon its members and their identity. In the case of material objects, one could argue that they are composed of and grounded and dependent upon their parts (and their identity). The identity of the extensional object is grounded in the formal collection of the identities of the grounding objects. 

In both cases, the composing and so grounding is simpliciter. Tuples – ordered lists of objects – where the order is important, and repetition is allowed – show this is not always the case. For example, <a, b, a> is a valid tuple where ‘a’ is a component twice. The tuple <b, a, b> has the same base components, a and b, but different orders (and repetitions) of them. In this case, the formal extension (and so criteria of identity) needs to recognise both order and repetition. (For more detail see Partridge (2019)) 

J.2.5 A solid basis for two types of extensional identity emerges 

Two development have changed the capability of extension to capture identity. 

The first related to the extension of material objects. René Descartes, in Principia Philosophiæ (1644), introduced the notion of material objects as res extensa – as extended in space. This notion took hold. For example, John Locke, in An Essay Concerning Human Understanding (1690), defined extension as "only the Space that lies between the Extremities of those solid coherent Parts" of a body. However, this raised questions about compenetration, where to two or more extensions occupying the same space at the same time – a problem raised by the Ancient Greeks. In the early 20th century, physicists developed a new way to think about extension, as space-time. Philosophers developed this as a more solid basis for identity. At one time, two things might be in the same place but not at a later time, having different spatio-temporal extensions. One could accommodate both (spatial) compresence and identity. 

The second related to the extension of general entities. It was normal, until the middle of the 20th century to regard the extension of general terms to be the actual objects that fell under the term. Work done by Kripke and then Lewis (see (Yagisawa, 2005)) on possible worlds provided an extensional way of capturing modal intensions (the extensions over possible worlds were sometimes called ‘intentions’) and so tracking identity. 

These two developments provided a solid basis for extensional identity. 

J.2.6 BFO example of universals’ non-extensional (intensional) identity criteria 

We have not been able to find any specific mention of identity criteria in the BFO manuals. However, the extract below (from the BFO manual) even though it does not mention identity criteria directly shows that it distinguishes between the universal and its extension in an intensional way. 

“2.6 Universals and classes 
Universals have instances, which in BFO are in every case particulars (entities located in specific regions of space and time). Universals also have extensions, which we can think of as collections of their instances. (Traditionally the extension of a concept is viewed is set-theoretical terms as the set of all the things that fall under the concept.) Such extensions fall outside the scope of this specification, but it is important for the understanding of BFO that the distinction is recognized. It implies further distinctions not only between universals and their extensions but also between universals and classes in general, including arbitrary classes such as: {the moon, Napoleon, redness}.” 

Basic Formal Ontology 2.0: SPECIFICATION AND USER’S GUIDE 
June 26, 2015 

J.2.7 OntoClean discussion of Intensional Identity 

Other ontologies address the question of intensional identity directly. The extract below is from the Wikipedia entry for OntoClean (“OntoClean – Identity,” n.d.) – a tool developed by the creators of DOLCE. This clearly talks about intentional criteria of identity – saying: 

“Identity criteria should be informative, they should help us and others understand what a class means.” It also offers two possible criteria for a triangle: it “… can be identified by the length of its three sides, or by two sides and an interior angle, etc. This says a lot about what is intended by the triangle class here, e.g. the same triangle could be in many places at the same time.” It seems to be relaxed that these are different objects as they have different definitions. 

Identity 

Identity is fundamental to ontology, and especially to information systems ontologies. Identity is well known in metaphysics and in database conceptual modeling. In the latter case, it is an accepted best practice to specify a primary key for rows in a table. If "two" rows have identical primary keys, they are considered the same row. 

More importantly for ontology are questions of identity that expose the existence of, or at least the need to represent, other entities. Here the issue at stake is finding the conditions under which a proposed entity would be both the same and different. The classic example is an amount of clay that is shaped into a statue. If you use the same clay but reshape it into a different statue, is it the same entity? If so, how could it be different? If not, how could it be the same. In conceptual modeling, it is understood that when such an ambiguity arises, one should treat it as two different entities to account for a situation where one changes and the other stays the same. 

In OntoClean, identity criteria are associated with, or carried by, some classes of entities, called sortals. A sortal is a class all of whose instances are identified in the same way. In information systems, these criteria are often extrinsic, like a social security number or universally unique id, which is not interesting from an ontological point of view. Identity criteria should be informative, they should help us and others understand what a class means. A triangle, for example, can be identified by the length of its three sides, or by two sides and an interior angle, etc. This says a lot about what is intended by the triangle class here, e.g. the same triangle could be in many places at the same time. Someone else may have an ontology in which the triangle class has different identity criteria, such that different drawings are always different triangles, even if they are the same size. Identity criteria (and OntoClean, for that matter) do not tell you that one of these definitions of triangle is right or wrong, just that they are different and thus that the classes are different. 

Identity criteria and sortals are intuitively meant to account for the linguistic habit of associating identity with certain classes. In the classical statue and clay example, we naturally say "the same clay" or "the same statue", indicating that there are identity criteria that are peculiar to each class. 

Being a sortal is the first OntoClean metaproperty, indicated with the +I superscript (-I for non-sortals) on a class in the original notation. +I (but not -I) is inherited down the class hierarchy, if a class is a sortal then all its subclasses are as well.

J.3 Indexicality 

In (Galton, 2018, pp. 37–8): "3.3. Indexicality: Past, Present, and Future: Should an ontology of time include reference to the present moment? As noted above, if ‘right now’ is seriously proposed, as in [1], as an example of a zero-dimensional temporal region, then the implicit answer is that it should. … If this is right, then it would be reasonable for a formal ontology that claims to provide an account of objective reality to include pastness, presentness, and futurity as attributes of times, though it is hard to see how to integrate this into the overall temporal framework, and as far as I am aware none of the currently existing formal upper ontologies has attempted to do this. Until such time as this is done, it would be best to steer clear of problematic entities such as ‘right now’ as examples of instants.” 

As I noted above and Galton does here, having a separatist spatial structure provides a good starting point for identifying the present. 

While Galton is right that many top-level ontologies do not have a way of representing the present, he is not right in that they all do not. There are top-level ontologies that include indexicality – BORO is an example. See Partridge, 1996: Business objects: re-engineering for re-use. Chapter 8 – Section – 4 – The time-based ‘consciousness’ of information systems – which discusses a ‘now’ and ‘here’ object. In a later paper (Partridge, 2018), a more sophisticated way of handling indexicality using agentology is described. This has already been discussed in the FDM forum –https://groups.google.com/g/uk-ndt-fdm. The paper suggests that there is an agentology layer indexed to the agent/system under the ontology. 

Return to Top

Return to Contents

Continue to next section: Appendix K - Glossary