Personal observations and understanding of realist applied ontology
The paper is about my observations and notes on applied ontology as systematic classification systems. The post explores what an ontology is, what is aimed for, how it shaped, what is its philosophical foundation, why this foundation is compatible with multiple philosophical positions. Inside the document, digression and further specifications and arguments are drawn to arrive at a deeper understanding of several multiple facets of the complex universe called ‘applied ontology.’ This is the second step toward the elaboration of a consistent ontology for my family enterprise, which will ultimately be used to ground its expansion in the market. Even though practically oriented, the observations can turn to be useful for all the researchers and scholars who are struggling to understand what an ontology is. That is to say to those who are not driven by a philosophical perspective (only). This is a working progress, and I will be ready to reply to any inquiring, comments, and questions from the interested reader.
Table of contents
- Introduction – 2. The general philosophy of applied ontology – iterative, non-definitive realism 2.1 – Ontologies, set theory, and truth conditions – 2.2 – Ontology as a way to representing reality 2.2.1 – A Kantian reply to the realist approach – 2.3 Reiterative principles of ontological representation as a way to construct practically an ontology 2.3.1 – Fallibilism – 2.3.2 – Fallibilism, scientism, and realism – 2.3.3 – An iterative process – 2.3.4 – Reusing principle – 2.3.5 – Fallibilism and Principle of Reuse – 220.127.116.11 – Inclusion incompleteness – 18.104.22.168 – Hierarchical and Structural Imperfections – 22.214.171.124 – Linguistic Limits – 2.3.6 – A vision on the long run – The strategic conception of a scientific realism metaphysical conception implemented by an epistemic fallibilist process compatible with the universal shareable representational artifact – 3. Ontology – Goals, Objectives, and Notions 3.1 – Ontology – Goals – 3.2 – Ontology – Objectives – 3.3 – Taxonomy & Hierarchy – 3.4 – Universals and Items – 3.4.1 – Universals – 3.4.2 – Items – 126.96.36.199 – Items inclusiveness in principle – 188.8.131.52 – How a general ontology would look like… to me! – [184.108.40.206.1 – Objects & Essential and Inessential Properties] [220.127.116.11.2 – Vacuum] [18.104.22.168.3 – Facts as Structured Objects] [22.214.171.124.4 – Events] [126.96.36.199.5 – Time and Space as Derivate Properties of Things] [188.8.131.52.6 – Histories] [184.108.40.206.7 – Beliefs and Social Objects] 3.5 – Relations among Entities – 3.5.1 – Relations among Universals – 3.5.1.a – 3.5.1.b – 3.5.1.c – 3.5.1.d – 3.5.2 – Relations among Items – 4. Conclusions
Since I started working on the ontological conundrums posed by the definition of war, that ultimately brought to a publication – The wild bunch is enrolled to the army – Sorites Paradox and the problems for the ontology of war (2020), I immediately realized how important is the field of applied ontology as systematic classification system for a couple of reasons. First, I needed to increase the relatively already efficient classification and storing system of my files and folds. So, I had to restructure it according to sound principles of classification to store and retrieve without losing time and… contents. The result was a set of methodologies to name, store and link different folds in different levels of what it can be rightly called “an ontology.” Sure, it does not follow a universal standard as BFO 1.0 & 2.0, but it is consistent with a sound hierarchical order where I can find everything I need without taking too much time to understand how I store something and why. Interestingly, such a systematic classification system can be studied from an empirical point of view simply counting the clicks needed to reach a given fold starting from a random point of the hierarchy (e.g. another fold), whatever their positions are on respect of the ontology they are part of.
In my case, all the folds are part of one unified system of folds. This system has at the top a super-class “GiangiuseppePili.” Inside it, there are the first classes – numbered 1.N where N is the number of each category, appropriately given considering how much I expect to use them. 1 is the most used, and 4 is the less used. in this way, I am always sure to get what I use the most at the top and what I use the less at the bottom. Among the 1.N categories, there is the “Super-Highway” category, which is a fold that contains only links to other classes. So, it is properly speaking an empty class, whose sole purpose is to collect links to other classes I use frequently, but they are positioned far below on the hierarchy. Indeed, I have 13.207 files stored in 1.705 folds, and, interestingly, the number of folders increases in the “bureaucratic area,” showing that the more you need to classify, the more you need to divide, distinguish and store. Indeed, even if my writing area(s) contains more files, it does not require a super-deep structure to be fully organized. This observation is also valid for the storage of other people’s papers (downloads). Indeed, to be fully efficient, a large and deep structure isn’t necessarily needed – it is a contextual matter to be solved.
The additions of links among categories were the major change in my ontology. The goal was to create a “little world net,” conceiving the hierarchy as a kind of net, with classes and items as hubs. Indeed, a hierarchy can be represented or mapped as a graph in which there are no natural ways to bridge the distance between far hubs (classes), especially if they are of different subsections. Interestingly, there is no clear law stating how many super-classes are needed to classify, map, and store all the items. I believe that this depends much more on the capacity of a mind to manage a substantial number of items than anything on reality – something that strife with some principles we will outline below.
The second reason that drove me to systematic classification systems was not due to any philosophical or organizational necessity. It was due to my renewed obsession with documents and certification. Basically, I finally realized that we live in a super-highly bureaucratically organized world where titles substituted human virtues. Bureaucracy is not only a set of practices to manage public state documents. It is also how an enterprise is organized through its set of documents. So, in a world economy driven by titles much more than sheer human power, ingenuity, and efficiency, mastering the difficult and divine art of documents is essential to be fully effective. The recent pandemic gave me a wonderful opportunity to play the war game of documentary systems against one of the staunchest bureaucratic state ever seen in centuries. Without entering too much in the details, the main point is already fully addressed in another post I had the pleasure of writing. The first and most important rule to remember is that a person – for a bureaucracy, is just a set of documents—nothing more and nothing less. The administration has workers whose role is to ‘read’ documents. This means that they will read a set of data put into a document. The bureaucrat will record how the data fit into the bureaucratic set of standards and rules. Therefore, if you want to get what you deserve from a bureaucracy, there is no other way than mastering the esoteric art of writing documents. A CV is exactly an instance of this kind. You can be the best organizer in the world, but if you don’t have a way to certify it bureaucratically, you are simply missing the chance to be classified according to what you have done.
As I stated in another place, it doesn’t matter the economic arrangement of the country in which you live. It can be a communist society or a libertarian capitalistic country. It doesn’t make any difference in the documentarian system. (Actually, it makes a big difference considering the effectiveness of the system as a whole and the power related to it. But this is not the place to understand the different impacts of documents between different economic systems). What varies is actually just the amount of rules and burdens the state put on you. But documents will be there in both extreme political arrangements. You are what you can certify, and you can certify exclusively what you are able to put into a form to be filled and approved by a third party, at least in principle – the CV is a good example of this specific case. So, I realized too late in life how important it is appropriately framing & certifying what you do. For this reason, my philosophical association elected me as the director of the diary and register – which is the secretary & historian of the association. Indeed, without appropriate expertise, everything done by the association would be lost. Capitalism and, even more, a highly bureaucratic state, is based on documents, and everything is set and inscribed into them.
This is the first step of an ambitious project, namely the ontological foundation of the family enterprise. The ultimate goal is to establish an entire structural and managerial infrastructure. The ontology as systematic classification system will be at the center of it. Then, the goal is to create the conditions to increase the relative added value of the firm from different perspectives, as outlined in the project already published. In addition, as the ontology is a systematic way to organize documents in a logical and consistent way, the goal is indeed to outline the managerial foundation for a franchising series of stores and shops. In order to create it, a highly effective systematic classification system is needed to create the capital. This is required to replicate the capital impact everywhere the documents are able to be enforced. This research was funded by Ancora dai Baruffi – Pizzeria Storica, inside the project “Improving your business awareness through philosophy – Structured Classification Analysis for Little Business Firms.”
2. The general philosophy of applied ontology – iterative, non-definitive realism
2.1 – Ontologies, set theory, and truth conditions – Interestingly, the literature on applied ontology as systematic classification systems is not united. The very conception of the goal and definition of what an ontology varies scholar by scholar consistently. For this reason, considering the development of Basic Formal Ontology (BFO) as an international standard, I would follow Barry Smith and colleagues’ spirit and ideas (Arper, Smith, Spear (2015). That is, I will follow the realist conception, which is – basically – the idea that the core mission of an ontology is indeed to represent a piece of reality through a system of constrained terminology and taxonomy grounded on rules to bound natural language. The concrete objective is to establish a structured taxonomy, where the ‘structure’ is intended to be a hierarchical order. In this respect, applied ontology as a systematic classification system is not a set theory variation for several reasons.
2.1.a. Set theory is ultimately an abstract theory to group things – which are, in theory, every possible object, but, in practice, it is ultimately about sets and particular kinds of sets.
2.1.b. Ontology is about representing reality as it is. Set theory is not about representing anything. Its goal is to define and characterize how we group things. Grouping is how we create sets and, possibly, numbers (mainly conceived as sets, though numbers offer a challenge to set theory – a topic that would require much more space than a sentence, but still an important point to be made).
2.1.c. For (1) & (2), ontologies include relationships among different entities whose nature could resemble some set theory operations. For instance, the sentence ‘x is_included_in Y,’ where x is a generic item and Y a generic category, is allowed by both ontologies and set theory. However, the ontology structure and logic are much more constrained in the definition of inclusion than for the set theory. So, stating that x is an element of Y in set theory can be just a trivial statement if that is the case. Instead, in ontology, stating that ‘x is an element of Y’ is indeed different because there is no such general thing as x and a general category as Y. This is because it must be checked that that is the case in reality, for instance, if a particular kind of dog is really an item of that particular dog race. To be true, the statement must include the definitions of what x is and what Y is, where the truth-value of the statement will ultimately depend on several conditions. Indeed they are not part of set theory as such and, again, because the set theory is not about representing reality where an applied structured ontology is. Then some conditions must be met:
2.1.I. Y must be a general category.
2.1.II. Y is something existing into reality (e.g., biological species).
2.1.III. The category can be of something existing potentially.
2.1.IV. X stands with for something existing.
2.1.V. x has all the features to be included in Y as its item.
2.2 – Ontology as a way of representing reality – In this conception, an ontology is about representing reality; therefore, it’s ultimately imbued by a philosophical realism. That is to say; the general notion is that (ordinary) language is able to capture a piece of reality as reality is. The realist conception elaborated by Barry Smith, Robert Arp, and Alan Spear expresses this thesis: “Realism: the goal of an ontology is to describe reality”. The statement is interesting, and non-trivial as far as it counts as a philosophical commitment to a kind of realism. The interesting question – for Smith et All., is why they should be entrenched in such a philosophical vision where this vision is indeed not necessarily required (see. 2.2.3). It is not entirely clear why an applied ontology whose ultimate aim is to ground a systematic classification system must be entrenched inside a specific philosophical framework. However, their commitment is not to a general realist approach. Indeed, it is in line with a kind of scientism, where they want to allow the classification of words as the science(s) define(s) them. Interestingly, their conception could be intended in two different ways, strictly and broadly. Broadly, because they ultimately set the stage to fix the jargon used into a specific domain through an appropriate ontology (for instance, see the work of William Mandrick and colleagues on military and intelligence domains (Mandrick, Smith et al. 2013a). Narrowly, because they ultimately look to science(s) specifically, their approach is indeed heavily scientific oriented. This is showed by the distance they took with a minimalist conception. The minimal assumption would have been just that language is indeed able to capture and represent reality as if we know perfectly how reality indeed is and whatever reality ultimately is. Scientism and realism could be considered as the proper philosophical framework in which basic formal ontology is grounded. We will return to this point below (2.2.3).
These philosophical commitments (scientism and realism) have both conceptual and practical consequences. First, practically, the goal is not to search for definitions and terms that mirror our conception of a field, discipline, or business. In this respect, the ontologist must definitely avoid defining terms circularly or differently from essential definitions (Seppälä, Ruttenberg & Smith 2017). Therefore, defining and identifying proper relations between categories and items/entities is about searching how reality is, using (ordinary) language coherently with the idea that it has to depict reality and not how we see reality. This brings us to the conceptual side.
Conceptually, along the line of realism, we cannot think the ontology is how we conceive reality. In fact, according to realists, it is perfectly the other way around. Ontology is and must be a representation of reality where ‘reality’ is intended to be something outside the mind, existing per se, and with a proper structure and set of substances. This is a neo-Aristotelian conception of ontology, where the effort is toward mirroring reality as it is and not as we conceive it. Then, knowledge of external reality depends on our mind. The problem is given by the distance between reality and the mind, where reality exceeds our mind. Indeed, there is no meaningful way to equate mind and reality. The struggle is posed by just striving to depict reality in its all details. Beyond this very possibility, a second assumption is needed, which is the capacity of the mind to mirroring reality as it is. To be clear, according to Arp, Smith, Spear (2015) the process of inscribing reality into an ontology is not a definitive act of the mind(s) but an iterative process that goes on through time improving progressively (see 2.3).
An ontology is not intended to be a definitive act of the mind translated into words able to maintain the relationship between mind and reality as the mind defines it – where reality is indeed prioritized epistemically vis a vis the mind and its eventual categories. Reality is at the center of the ontological process, though this process starts because of the need of fixing ordinary languages in particularly forms by a group of interested subjects (with a rational mind and linguistically capable). Then, even if ontology in this sense has reality at its center, it is ultimately the intentional goal defined by the mind which eventually starts and drives the process itself. It is a progressive activity where reality is consistently reconstructed (and not constructed ex novo) by the ontologist. Then, these are philosophical assumptions from a realist point of view:
2.2.a Reality is a structured set of things.
2.2.b (2.2.a) exists independently by any mind.
2.2.c The mind is able to represent reality for what reality actually is.
2.2.d The process of describing reality for what it is, is an open-ended and fallible
2.2.e A particular process to represent reality into language is an ontology
2.2.f An ontology is able to represent reality for what actually is.
2.2.g A constrained language can be shared if the set of constraining rules is given and shared.
2.2.h As a set of rules to constrain language, an ontology can be shared.
Is the realist approach to ontology also committed to a correspondent theory of truth? Is it constrained to truth at all? I would say that probably is not constrained to the correspondent theory of truth, but it is certainly constrained to a notion of truth that is indeed a realist one. It seems that the correspondent theory of truth is the best one for capturing what an ontologist is trying to do. However, this general purpose can be reached because language is able to capture reality not randomly, when this is the goal. In this sense, realism is only constrained to the commitment that there is a way to use the language that is able to depict reality. Let’s consider two possible interpretations of the use of language in the light of what was just stated for the realist approach. Considering Wittgenstein’s positions, there is no either/or choice between the Tractatus language and the late Wittgenstein’s conception (Wittgenstein 1921, 1953, Pili 2017). The former conceives logics as the grounding language, and the ordinary language is its imperfect derivative. The latter considers logics as a particular variation of all the possibilities allowed by ordinary language. In this second conception, ordinary language is prior to logics logically and causally. So, considering the ontological approach we outlined, ontological realism is compatible with both, and it endorses a perspectivalism conception toward descriptions. It states that “there are multiple accurate descriptions of reality”.
2.2.1 – A Kantian reply to the realist approach – The theoretical realist commitments to ontology seems to lead inevitably to a rejection of Kantianism. Smith explicitly considers this option in his lectures, at least in the strict interpretation of Kant’s categories in which they constitute our conception of reality. However, I would argue that an epistemological Kantian – as myself – is perfectly fine with the realist conception of (applied) ontology. Indeed, what a Kantian is only committed, and he/she must recognize, is that the mind is not free in describing reality, because its own categories rule its judgment once and for all. First, a Kantian is neither skeptical nor idealist. And, according to what I had understood of the Kant’s epistemology – carefully avoiding his commentators, a Kantian approach to reality can be closer to a realist account than to an idealist one. This is how I see the Kantian project. It consists in limiting the statements of what exists to what can be actually put into a map. This is compatible with (2.2.a-f). However, the limit of knowledge of the realm is fixed by experience. We can state x exists if and only if a subject S has sense-data of x collected together by the intellect into an object. Then, there is no way to state x exists in the world besides what you can ascertain through empirical knowledge. This version of Kant’s epistemology is compatible with realism in a sense intended in these notes. But even considering a different Kantian conception, it can be a useful working principle stating that you try to represent reality as if there is one, structured and independent reality. The only fictional part is to state that reality is “structured” because Kant’s noumenon is indeed “independent from the subject” and it is not produced by the subject. This fiction could be accepted readily by a Kantian – whatever its epistemological interpretation of Kant’s philosophy will be. What cannot be asked to a Kantian is to conceive this representation as reality necessarily is. This means a Kantian cannot accept the idea that reality is as it appears to us. It can be, but there is no way to ascertain it. To us, there is no other way to conceive reality, but what reality is without us, cannot be determined once and for all. But this limit is, in a sense, endorsed even by the realist. The development of an ontology is not a straightforward process, fixed once and for all, as we shall see in 2.3. In the progressive conception of reiterative selection and learning of how reality really is, there is a sense in which reality is a given but not entirely dominated immediately by the mind. This is how science indeed works, and then as a particular kind of scientism, this is how they portray the quest of ‘truthful ontology’. Insofar it will never be a definitive conception of a one-to-one representation of reality; an epistemic Kantian is happy and fine with it. Of course, an epistemic Kantian, close to a realist variation of Kant’s philosophy, is even happier and fine in comparison with the idealist interpreters of Kant. They are the only ones who can start to complain about this realist approach. After all, this realist conception looks like a normative methodological aspiration than anything more than that.
2.3 The reiterative principle of ontological representation as a way to construct practically an ontology
2.3.1 – Fallibilism – Coming closer to the process of ontological representation, there are further meta-methodological principles outlined by Arp, Smith and Spear to be considered. These principles will show further the logic of the ontology construction. First, it is a fallible process. This was already anticipated, and it is a logical consequence of scientism and, in a lesser extent, of realism. Indeed, as far as science itself is sensible to time evolution because (a) the main theories that give meaning to scientific statements can change. (b) It includes more and more sense data through experiments. Therefore, for (a) & (b) it can be the case that science uncovers more items and categories (as the history of biology shows perfectly – for instance see by Charles Darwin and Richard Dawkins) to be added to the scientific realm of entities. Another clear example is the standard model of physical particles. For a while, particles were discovered and added to the standard model. Then, an applied ontology committed to scientism is also committed to fallibilism, something that Karl Popper would have strongly appreciated.
2.3.2 – Fallibilism, scientism, and realism – Interestingly, if there is no limit on the epistemological process of reality description encapsulated into an ontology, the limit of fallibilism is given not by scientism but by realism. The process of discovery can be infinite, but there is still one and only one reality to be depicted. Therefore, if a piece of reality is exhausted appropriately by an ontology, then it is exhausted once and for all. Then, a distinction here has to be made. From one side, fallibilism and scientism allow an endless epistemic process translated into an ontology, but they have a metaphysical limit in what can be portrayed as far as reality is indeed the extreme limit of the process itself. In theory, reality can be exhausted by a fallible process of description. In practice, it will never be the case that we can be sure to have exhausted the piece of reality we wanted depicted. However, considering the scientist side of the realism proposed here, what science considers must be taken as the ultimate reality as far as we know it at a given time.
2.3.3 – An iterative process – 2.3.2 shows what is allowed in theory. In practice, the process of constructing an ontology is an iterative process in which the ontologist structures the ordinary language, fixes it into a set of core definitions, puts the terms into a hierarchy shaped by a consistent set of specific relations (inclusion, partition, etc.), and possibly writes it into an appropriate program. The most interesting part of the process consists in the necessary interaction between the ontologist and the experts of a given field at least potentially. An ontologist – unless he/she is also an expert of the specific portion of reality he/she is going to consider – must always undertake a process of scientific discovery of the specific agreed common uses of the terms in a particular scientific domain. Indeed, he/she would have his/her own uses of the words, but he/she must be very careful in fixing the definitions and putting them into a hierarchy. Namely, he/she has to avoid idiosyncratic definitions and personal uses of the terms as much as possible. This can be done only considering how the scientific community of a given field actually uses the terms and how it implicitly structures them into their ontology. Then, the process is constrained to an iterative epistemic investigation in the actual uses of the terms. Therefore, the ontologist must always be careful about avoiding his/her own conceptions of terms and concepts on which discipline is grounded. For my own experience, I discovered how differently people perceive and use the same words. They are always slightly differently interpreted person by person, given their different grasp of the discipline as such. It can be a reward – or a frustrating – process to go through because, interestingly, the differences of the uses can be subtle but substantial. The ontologist has to be careful and aware of this, and he/she has the specific duty to represent the use of the language as it is used by the community and not by him/herself.
2.3.4 – Reusing principle – An ontology is an informational artifact. An informational artifact ontology was explored by Smith and Werner Ceuster. Specifically, an ontology is thought to be a representational artifact, comparable to any kind of artifacts. That is to say, an ontology – as already stated (2.2), has a specific goal that is to represent reality as reality is – whatever reality would turn to be – through constraints to the uses of ordinary language. Then, an ontology as systematic classification system to be employed by several people cannot be thought to be un-reusable. After all, it is not thought to be an idealistic self-portrayed representation of reality. Therefore, a good ontology is necessarily re-usable. As Arp, Smith, and Spear framed it into a specific principle: “The principle of reuse: existing ontologies should be treated as benchmarks and reused whenever possible in building ontologies for new domains”. Then, this is practically essential for any winning ontology, that is its ability to be (a) reusable by any person involved in the field but (b) without giving up with the realist principle (2.2). The reusability principle is of course compatible with the scientism already defended (2.2), and with the fallibility principle (2.3.1 & 2.3.2).
2.3.5 – Fallibilism and Principle of Reuse – Fallibilism doesn’t entail the reusability principle, but it merges with it naturally. Indeed, the reusability principle allows changes from other users without hampering all the ordinary process of building the ontology – if the ontology is indeed well structured so to meet this possibility and requirement. In this way, in a group working activity whose goal is to establish an ontology, different users can signal mistakes, errors or insert items not previously considered. This means that an ontology, as an information artifact, is able to adapt quickly and efficiently in two different senses. Indeed, an ontology can be flawed by several mistakes but mainly they are reducible into three categories (220.127.116.11, 18.104.22.168, 22.214.171.124). The first category includes incompleteness and inclusion; the second category includes hierarchical and structural imperfections; the third considers linguistical issues.
126.96.36.199 – Inclusion incompleteness – The ontology does not include all the items it should consider. Then, the ontology can be complete – that is to say, an ontology can put all the items into the proper hierarchical structure, but it can miss exhausting all the items in the field. Then, for instance, if a new basic physical particle is discovered, it must be named and, as such, added to the previous ontology. The ontology could have been perfect hierarchically, structurally, and logically but it is committed to expanding its set of items. Therefore, if the ontology does not exhaust all the items, an ontologist must add the missing item as soon as possible. However, this can be done if the ontology is constructed in a way to easily fix this issue. If the ontology is not appropriately planned to be flexible, this solution could not be available. This underlines the most important issue, which is the logical consistency and adaptable structure of an ontology.
188.8.131.52 – Hierarchical and Structural Imperfections – The ontology can be flawed structurally such that it doesn’t allow any further change or progress. This means that any change needed could turn to be an ontology killer or, at least, it could require a sensible restructure on the ontology as a whole. Mistakes of these kinds are determined by a flaw in the first conception of the ontology as such, ossified into the wrong hierarchical organization. For this reason, the fallibilism approach is so important. Considering that usually an ontology is a collective enterprise, and therefore it must be used also by other people than the ontology him/herself, it is particularly important that its structure is indeed compatible to an organic change. So, if a subpart of the ontology is flawed, it should be adjustable without hampering all the other subparts of the ontology itself.
184.108.40.206 – Linguistic Limits – Language is in between the ontology as a representation artifact independent by its verbal formulation, and the reality that is considered and inscribed inside it. Then, language is crucial as much as a bridge in connecting two different sides of a street divided by a stream. In addition, the appropriate linguistic formulation of an ontology is a critical part of the ontology as ‘infrastructure’. Therefore, if the ontology is poorly written, it cannot be fully adaptable or usable. Considering that the ontology is indeed mainly a taxonomy put inside a hierarchy, its adaptability is an essential feature of winning ontologies. Therefore, an ontology must be consistently planned to include a fallibilist expansion in two different directions: (1) improvements over the items to be included; (2) agile ability over changes on subparts reform where needed.
2.3.6 – A vision on the long run – The strategic conception of a scientific realist metaphysical conception implemented by an epistemic fallibilist process compatible with the universal shareable representational artifact – Closing this section, it is important to stress that the right approach in planning and conceiving an ontology is, indeed, looking to the long run. Again, this is because the ontology goal is primarily to represent a portion of reality as it is. However, given the relative ignorance we have on (a) how reality ultimately is, and (b) how the ontology will be ultimately used by the users, it is substantially important to look to the long run. Indeed, a fallibilist approach works only if we allow it to go through time. Time is an essential variable in all the domains in which a process runs and selects what works from what doesn’t work. In a sense, the ontology resembles an animal that organically adapts to its own environment. Any kind of selection, natural or artificial, cannot be forced into one time shot. It requires time to check actual mistakes, possible revisions, and amendments. The process requires a natural and artificial pressure, friction of war. Perfection is never a given, always a result. Therefore, a sound strategy in planning an ontology must always allow such progressive adjustments considering the three more or less unavoidable mistakes (see 220.127.116.11 – 18.104.22.168).
3. Ontology – Goals, Objectives, and Notions
3.1 – Ontology – Goals – The ontological goal is to represent a portion of reality as it is through a general constraint of the ordinary language – intended as the way of a given community uses its own language (in this specific sense I am considering ‘ordinary’ such language). This is the goal of any ontology.
3.2 – Ontology – Objectives – The intermediate objectives to be achieved depends on the definition of the domain to be ontologized, its scope, and dimension. There are different kinds of ontology from very general ontologies to very specific ones. This depends ultimately on the mission assumed intentionally by who is going to structure and define the ontology itself.
3.3 – Taxonomy & Hierarchy – As far as an ontology is based on specific constraints on natural language in order to represent reality through a systematic hierarchical taxonomy, it is essential to grasp how an ontology is at the end. Then, first, the realist approach assumes that reality is indeed organized hierarchically. That is a logical consequence from the realist approach itself, because if ontology is able to represent a piece of reality and if an ontology is essentially a hierarchical structure of well define terms, then to be an ontology meaningful it must be assumed that reality itself is hierarchically organized. Otherwise we would have the paradox of having a trustful representation which is not as reality is (hierarchical). This would be a strong inconsistency inside the realist ontological account. Then, we must assume as a given in the framework of the realist account, that reality itself is structured and indeed the ontology mirrors reality in this essential feature. So, this implies that looking how an ontology is, is a way to see how reality is essentially, i.e. at its core.
An ontology is a set of definitions of terms whose organization is inscribed into a hierarchical, consistent, and logical structure. Then, first, an ontology is domain oriented and committed to it. This means that it must include all the items of that given domain. Second, the ontology has to organize its set of elements into a set of definitions of all the terms that identify them. These terms must be regimented into a set of categories that includes those items inside them properly. This means a realist conception of categories. Then, a category is like a specie and an item is a specific animal. That is to say that a category essentially captures what is part of it. Then, a hierarchy of categories is what really shapes the structure of an ontology. For this reason, interestingly, even though the ultimate goal of an ontology could be to systematic classify items, it’s only through the definitions of categories, their subdivisions, partitions and inclusions that ultimately constitute the ontology and its real value.
3.4 – Universals and Items – There are essentially two possible entities in an ontology, universals and items.
3.4.1 – Universals – A universal is an essential category of things or entities. It is a set whose definition includes several essential features. Then, it is a class in the sense in which a species or an animal race is a class, that is it captures something essential of what is indeed included into it. Instead, an item is generically what can be captured by a universal. Therefore, if there is an item x then there is a universal Y that captures x. In ontology, there is no overlapping between universals. This is a fundamental difference between set theory and ontologies as far as set theories allow near infinite overlappings between sets (one element x that belongs to a set X can belong potentially to a in infinite numbers of sets W, Y, … Z). In fact, if every item is indeed defined by a set of properties, if each property is a class inscribed hierarchically into an ontology, if each property distinguishes different essential features, then an item is included into a set of structured categories. Then, the lowest category into an ontology can be part of other classes as well but it is basically entrenched into those categories in a way that its function is to identify more specifically a (set of) items. Then, if a general category is ‘animal’, a subclass is ‘fishes’ a more specific category is ‘shark’ and finally it ends with specific subclasses of shark races (e.g. ‘white shark’). This is in practice how an ontology works, that is identifying general categories to be furtherly divided in a way that the subparts do not overlap between them – e.g. there are only essential partition sets-categories. Again, the partition operation over a single set do not necessarily lead to a meaningful subdivision from an ontological perspective. The ontological partition is not just a random subdivision. Indeed, the partition is ontologically meaningful only if the subcategories identified are also capturing an essential difference, a substantial essential feature of reality, namely they distinguish different items looking to their essence. A classic random partition of a set is not interesting ontologically because it is not about capturing essential features of the elements divided into different sets. So, to sum up, a universal is a class that identifies essential features of what exists in reality. A universal can be naturally divided in two or more subcategories only if those subcategories are identified by a distinctive essential feature that belongs only to a subset of the entities included into the more general category. Then, a logic of inclusion is such that an upper category includes all the elements of the subcategories inscribed into it.
3.4.2 – Items – The applied ontology, as systematic classification systems, is planned to include everything existing in its own distinction. That is to say that there is no unifying principle or a unifying mission. What is important is the ontology’s ability to depict reality as it is, not to find the ultimate solution of the ultimate constituent of reality as such. In this very sense, it is far from being a philosophical approach, and that’s why we discussed the philosophical assumptions directly in the beginning (2). There is no further commitment apart from realism and scientism in a very broad conception of them. Then, an item is just a specific entity captured by the ontology, and it must be inscribed into the hierarchical order of classes and subclasses. As a general rule, an ontology must be inclusive as much as possible, namely an ontology must be planned consistently to include all the elements inscribed into its domain. Then, if an item is excluded, and if its exclusion is discovered, the item must be included into the ontology.
22.214.171.124 – Inclusiveness in principle – The realist ontology advocated here is neutral from a philosophical perspective on what is substantially essential and able to ground everything else. For instance, the atomistic conception stated that atoms and vacuum are the two ultimate entities of the world because the aggregation and motion of atoms generate everything existing, which is ultimate a very complicated – not complex – thing. This is a way to reduce everything existing to something more fundamental, and this is a long-lasting philosophical goal. That’s why applied ontology is not a philosophical enterprise. Therefore, applied ontology is only committed to represent reality, not to find the ultimate component(s) of it. This means that the goal is just to include items into a hierarchical order that, allegedly, is mirroring how that particular piece of reality is structured. There is no research on essential components of reality. For this reason, applied ontology of this sort considers a wild variety of entities inside it: structures of objects, properties, events, processes, and relations.
126.96.36.199 – How a general ontology would look like… to me! – Considering 188.8.131.52, I want to propose my own version of metaphysical conception of the world. Ultimately it is a form of Spinozism and then compatible with a realist conception. I am not going to elaborate or defend in detail my position, but I couldn’t resist the temptation to report it here. This is not the place to a full defense, but I take the chance to apply it to intelligence and war. The interested reader can already explore pieces of it on my YouTube presentation of ontology of war and on my recent paper The wild bunch is enrolled to the army – Sorites paradox and the problems for the ontology of war. On my book Filosofia pura della guerra, I deliver a metaphysical conception of war which will be soon presented to the international reader as well. [184.108.40.206.1 – Objects & Essential and Inessential Properties] So, coming directly to the point, the world is divided into different objects, whose nature is identified by sets of properties. Those properties can be divided into essential properties and inessential properties. The essential properties are those able to identify the objects grouping them in universals. Universals are not existing into reality, but they are a proper way to group existing things. Instead, the inessential properties are those which distinguish each single object. Then, the Leibnizian principle of the identity of the undiscernible entities applies exactly because there are the inessential properties – where the location is in space and time can be considered an inessential property of an object, though really an important one. The essential properties uniquely identify universals, and universals capture single things. Space and time are just ways to conceive objects and, then, they do not exist without objects and minds that actually conceive those objects. [220.127.116.11.2 – Vacuum] then the vacuum is not an object as such, and it is excluded by my metaphysical conception. The vacuum is only constructed by a mind as a limit notion created by opposition, comparison and, ultimately, by subtraction. After all, there is no way to define it and to identify anything existing with the definition of vacuum proposed. Then, at least, if vacuum exists, its existence cannot be proved. In a similar fashion, death exists in the sense that it is a limit of our ordinary conception of the world, which is ultimately unthinkable. As the universe as a totality is just a confused though important idea, the same can be said of death. It indicates a limit of our ordinary experience, the place where we cannot go on. Nothing more but nothing less. [18.104.22.168.3 – Facts as Structured Objects] Combinations of objects generate facts only if those objects are indeed put into an order, which is ultimately a structure. The existence of an object is independent of space and time, at least there is a sense in which their existence is time and space independent. Logically, their definition is not time sensitive. After all, everything existing in time is a process and a process is not an object, but it requires objects to exist. We shouldn’t be deceived by this ultimate reality, namely the time-independent nature of objects. For instance, a person is a process that goes on through time. Then, structured sets of objects are facts. [22.214.171.124.4 – Events] Events are combinations of facts put into a net of relations. The kinds of relations define the different natures of events. Causal relations define existing events in the world as natural facts. Informational relations define existing events in the minds and, among them, of abstract entities which are, according to me, just expanded figments of our minds constrained by general rules. There is no such thing as the number 1, apart what we define, extrapolate and extend as such, giving it a name and using it into our language as if it indicates an existing object independent from our minds. [126.96.36.199.5 – Time and Space as Derivate Properties of Things] Then, if facts are ordered sets of objects, and if events are ordered sets of facts related by specific kinds of relations, events exist in time in the sense that we conceive them as such. Causality or informational proximity generates the perception of the existence of time as such. However, my long-lasting rejection of the existence of time as an independent component of reality that would exist even in a blank universe is based on the idea that there is no way to identify what a time sequence is without using objects, considering the mind as an object, that is to say, we always need a clock. Without clocks – however, they are made, there is no way to think time. So, time is just a way we compare different things we perceive in a given order. That given order where one object comes first and a second object comes second, that is time. Then, time emerges once we had introduced the notions of (different kinds of) relations. Picking up two different things, putting them into a structure, and, basically, into a relationship that is how time emerges. Without this ability, which is ultimately an ability to reduce reality into a set of universal rules, time, and space wouldn’t be conceivable. This is the argument for considering time as a matter of the relationship between different things, where time is ultimately defined by a clock in a very strict sense. Space, instead, can be just seen as an abstraction of portion of things cut into pieces or considered as a whole but distinguished by other things. In this respect, again, there is no sense to speak about space without a previous abstraction and further generalization over that abstraction. Actually, this conception of time and space is indeed quite attuned to epistemological Kantianism to which I agree completely where it comes to metaphysics and the limit of reasons. Processes are structured sets of events. [188.8.131.52.6 – Histories] An history is just a set of events in which the events are collocated into a time order. It could be thought that set of histories could lead ultimately to totality. However, as already stated, totality is just an important limited concept of our understanding. Thus, my argument here is just that sets of histories always lead to other histories and so on and so forth. Basically, there is no meaningful way to combine histories in order to achieve a complete totality. It is always possible to add a new story, a new set of events to a previous one. That said, again, if there is a metaphysical universal conception that still I appreciate in its beauty and power is Spinoza’s conception of universal totality as absolute infinite nature. [184.108.40.206.7 – Beliefs and Social Objects] Social objects are the result of a mix between different kinds of entities: physical objects, information, and beliefs. Objects were already defined [see 220.127.116.11.1]. Physical objects are those defined by causal-effect relations of the natural kind. Informational objects are those defined by pieces of information inside computations, which are basically formal transformation through rules. Beliefs are states of minds whose existence is identified by their belonging to a mind, they are not causally influenced directly by other kinds of natural facts, and they do not influence directly other kinds of natural facts. As objects, their nature is fixed by informational qualities such as clarity and distinctness, accessibility etc. Social objects are obtained by a complex interaction between physical objects, and mental objects through language. The central pivot is indeed language itself, which is needed to fix mental and informational objects through a set of shared rules. Social objects isn’t possible without a sort of agreement among different minds that decide to converge over a constrained behavior fixed by rules defined through language. Language is a means and an end. As a means, it is the way through which the minds define rules of behavior to constrain themselves when they translate themselves into physical events through appropriate causal relations (mind-world). Then, two or more different minds communicate linguistically – simplifying the chain of rational reconstruction – first, to be sure they use the language in a compatible way before the agreement. Second, they fixed the rules to define a common field of meanings for their own language. Third, they introduce and define their rules. Finally, they use those rules to define laws of interpretation (language) and behavior (events). Of course, this is not necessarily done explicitly and consciously, but this is a logical and plausible reconstruction of the process. Then, after an agreement is reached to fix a penumbra of common meanings, a set of rules is now possible and they constitute the ground for defining the laws of behavior and further linguistic constraints, interpretations, and terminology. This is indeed how an ontology is basically made, as an agreement over the constrained use of the language. In this sense, an ontology is a social object as far as it is made by minds constraining language through rules to define common meanings and rules. So, a social role is an agreement among minds to interpret a physical item – a person, to be counted as X in the context Y, where the context Y is a field of meanings in which the role is recognized by the community who defined the social role. An informational artifact is basically similar to a social role with the difference that it is implemented by a physical object to be used or computed by another informational artifact that ultimately is used by a mind.
3.5 – Relations among Entities – I beg the pardon of the reader for the long philosophical digression, but it was interesting and too tempting avoiding how a general philosophical ontology can look like, in order to be more or less compatible with a realist account from a neo-Spinozian & Kantian perspectives. Coming back to the core notions of applied ontology, the last paragraph is dedicated to relations among entities. There are three different general relationships: universal/universal, universal/item, item/item. Surely, this is not the place to exhaust the topic, but it is worthy at least to consider what is really essential to be grasped.
3.5.1 – Relations among Universals – The relations between universals are basically of four types, and they are defined by the hierarchical position between them: (3.5.1.a) a universal A is up to a universal B, if B is directly included into A, then B is a subclass of A and B is a universal in its own respect. So, if A is “mammal”, B is “swimming mammal”. As it is clear, B (3.5.1.b) a universal A is at the same level of B and A and B are both subclasses of the same progenitor C, then A and B are in a relation of proximity. In this respect, A and B identify diversity in essential properties among the same universal. So, for instance, if C is “mammals”, A can be swimming mammals and B earthly mammals. So, A and B identify both two distinct categories of the same class (mammals). As it is clear, 3.5.1.b is a particular case of 3.5.1.a. Indeed, it is simply the case in which the progenitor of B is A itself. It is a borderline case of 3.5.1.a but still it is a valid one. So, a similar point can be made for the next subdivision, 3.5.1.c. (3.5.1.c) Two universals A and B can be both subcategories of the same progenitor C in a different level, for instance, A is “dog” and “B” is “sea mammal”. They both are part of the same universal “mammal” but they are positioned in different points of the hierarchy and in different partitions of it. (3.5.1.d) Two universals A and B can be put in completely different categories because they simply are of different substantial kinds, such as fishes and helicopters. (3.5.1) is not thought to be an exhaustive account of relations among universals but still it gives an essential picture of how an ontology is shaped and organized among subdivisions and partitions of universals. Interestingly, let’s just consider for a brief moment an important fact. Again, the subdivision and partition of subcategories in an ontology is not a set theory partition operation. Indeed, interestingly, the subdivision among universals with the same progenitor can be a partition set, but this is neither necessary nor the rule. It is indeed only by chance that the subdivision of a universal generates a partition set. This is because they are mainly two different operations. A set partition is just a way to divide a given set in a way that all its elements are subgrouped inside the given set. Instead, the subdivision of a universal is not aimed to exhaust all the possible subsets that can be created inside of it. It is aimed just to identify different essential properties of fundamental distinct items inside of itself. So, there is no need to subdivide a class if that class is already sufficiently able to identify what the ontologist thinks is a basic entity of its ontology. For instance, in an ontology of pizzas, it can make sense to subdivide pizzas into their components or it can be a useless operation. It depends on the goal level of that precise ontology. For a business restaurant ontology, it is meaningless to subdivide things coming to the known atomic subparticle. It does much more sense to stop at the ingredients level.
3.5.2 – Relations among Items – The relations among items is similar to the relations among universal and it is basically reducible to the relation of belonging from an item to a class. I don’t think it is necessary to go through this section to grasp what are the relations among items.
This article was not aimed to exhaust or even to portray fully what applied ontology as systematic classification systems are. I thought to take organized notes for myself hoping they can turn to be useful to mark a point also for other people that struggled as I did on understanding what basic formal ontology is about and its possible applications and variations. I strongly recommend reading all the papers in the bibliography and possibly having a look at Barry Smith’s YouTube channel, where there are courses and impressive videos from one of the world-leading ontologists. As it was stated through and through this post, this analysis is consistent with a realist approach to ontology, which is compatible with user revisability and scientism. It was argued that this approach is also compatible with a neo-Kantian philosophical conception, and it is only difficult to be accepted by neo-idealistic philosophical positions. I want to close this short analysis stating that applied ontologies as systematic classification systems are much more a practical endeavor than philosophical research. The post focused more in the latter aspect just because it was not the right place to show how to make an ontology practically. That said, it is crucial to work through a trial and error process, experimenting, and playing a bit to understand how it can be done. It is not easy making a good ontology, and when it is faced with the problem for the first time it must be taken seriously but using the right ingenuity to figure how to do it at best knowing that – maybe – it must be done over and over again!
- De Soto, H., (2002), The Mystery of Capitalism, New York: Basic Books.
- Arp, R, Smith, BD & Spear, AD (2015), Building ontologies with basic formal ontology, The Mit Press, Cambridge Mass.
- Pili, G., (2020), “The mystery of capitalism is doing things with documents”, scuolafilosofica.com, https://www.scuolafilosofica.com/8089/the-mystery-of-capitalism-is-doing-things-with-documents
- Křemen, P, Mička, P, Blaško, M & šmíd, M 2012, ‘Ontology-driven mindmapping’, Proceedings of the 8th International Conference on Semantic Systems – I-SEMANTICS ’12.
- Seppälä, S, Ruttenberg, A & Smith, B 2017, ‘Guidelines for writing definitions in ontologies Selja Seppälä’, Ciência da Informação, vol. 46, no. 1, pp. 73–88.
- Slimani, T n.d., A Study Investigating Typical Concepts and Guidelines for Ontology Building.
- Smith, B 2012, ‘How to Do Things with Documents’, Rivista di estetica, no. 50, pp. 179–198, viewed 10 March 2020, <https://journals.openedition.org/estetica/1480>.
- ― 2019a, BFO Tutorial (2019). Part 2: Why Ontologies Fail, YouTube, viewed 12 March 2020, <https://www.youtube.com/watch?v=w5d5KmBqw3w&list=PLyngZgIl3WTj6tWcypTLpCnYXu6o93kD4&index=2>.
- ― 2019b, BFO Tutorial (2019). Part 3: Qualities, Dispositions, Diseases, YouTube, viewed 12 March 2020, <https://www.youtube.com/watch?v=2UmKWQ-fH4s&list=PLyngZgIl3WTj6tWcypTLpCnYXu6o93kD4&index=3>.
- ― 2020, BFO Tutorial (2019). Part 1: Introduction to BFO ISO, YouTube, viewed 12 March 2020, <https://www.youtube.com/watch?v=muafRW0bXgw&list=PLyngZgIl3WTj6tWcypTLpCnYXu6o93kD4&index=1>.
- Smith, B & Ceusters, W 2015, “Aboutness: Towards Foundations for the Information Artifact Ontology”.
- Smith, B, Malyuta, T, Rudnicki, R, Mandrick, W & Salmen, D 2013a, IAO-Intel “An Ontology of Information Artifacts in the Intelligence Domain”.
- ― 2013b, IAO-Intel An Ontology of Information Artifacts in the Intelligence Domain.