Saturday, June 13, 2009

Clouds of ontology, Part 1

The last few years I have been encountering expressions containing the word "ontology" in the context of natural sciences, computer science and the internet - "medical ontology", "web ontology". For some reasons or other I feel deeply suspicious about all this, without yet knowing zilch about it. I suppose one reason is that it appears to involve a lot of freshly-minted software, fancy terminology, amiably intelligent hotshots and research loads-a-money. After 25 years in IT, I am fed up to the teeth with that - except for the last item, natch.

I should just mention that I find "ontology" and "epistemology" to be fairly useless words. They made serious sense only in the context of the dogmatically dualistic world-view sometimes called "Cartesian". They have appeared in various philosophical guises in the past, for instance in connection with the similarities and differences between "Venus" and "the morning/evening star". Supposedly someone asked Tolstoi what the difference was between governmental violence and revolutionary violence. He replied, "the difference between cat shit and dog shit". Of course that may be somewhat unfair. I'm sure it makes a difference to upholstery cleaners.

Anyway, I have put on my red riding hood, grabbed my basket, and gone out zilch-collecting in Entity Forest. So far, I have found a few possibly digestible items, and only two toadstools - but these two are quite toxic. I will describe them in a subsequent post. But now, a word about our sponsor: "ontology". (Actually not yet a sponsor, but I'm going to apply for a grant).

One of the success stories in the ontology business appears to be Barry Smith. I reported slightly sarcastically on his having won big money 10 years ago. He is professor of philosophy at the University of Buffalo, and editor of the revived Monist. I once watched some of the videos in a training course he did called "Introduction to Biomedical Ontologies". They're extremely interesting and well-presented, in contrast with other stuff I've found on the internet. From Smith's course, a picture is emerging for me of serious and useful work being done at least in medicine and genetic science. But attendant on this are gigantic clouds of brow-knitted philosophastering - as has ever been the case in IT, where those involved are called consultants.

The guy credited with popularizing the term "ontology" in its new sense in 1992, beyond the confines of AI, is Tom Gruber. He originally wrote: "An ontology is a specification of a conceptualization". Hmmm. But that was only the short answer. He goes on:

What is an Ontology?

Short answer:
An ontology is a specification of a conceptualization.
The word "ontology" seems to generate a lot of controversy in discussions about AI. It has a long history in philosophy, in which it refers to the subject of existence. It is also often confused with epistemology, which is about knowledge and knowing.

In the context of knowledge sharing, I use the term ontology to mean a specification of a conceptualization. That is, an ontology is a description (like a formal specification of a program) of the concepts and relationships that can exist for an agent or a community of agents. This definition is consistent with the usage of ontology as set-of-concept-definitions, but more general. And it is certainly a different sense of the word than its use in philosophy.

What is important is what an ontology is for. My colleagues and I have been designing ontologies for the purpose of enabling knowledge sharing and reuse. In that context, an ontology is a specification used for making ontological commitments. The formal definition of ontological commitment is given below. For pragmetic reasons, we choose to write an ontology as a set of definitions of formal vocabulary. Although this isn't the only way to specify a conceptualization, it has some nice properties for knowledge sharing among AI software (e.g., semantics independent of reader and context). Practically, an ontological commitment is an agreement to use a vocabulary (i.e., ask queries and make assertions) in a way that is consistent (but not complete) with respect to the theory specified by an ontology. We build agents that commit to ontologies. We design ontologies so we can share knowledge with and among these agents.
"Semantics independent of reader and context": that's a nice one! In his philosophical reading, Gruber possibly didn't get as far as Gadamer and Luhmann, to name but two. I bet even old Schleiermacher would have dropped his veil in shock. It's also a tiny bit inconsistent to say ontology is "often confused with epistemology, which is about knowledge and knowing", and yet claim that his ontologies are "for the purpose of enabling knowledge sharing and reuse". If an ontology is a prerequisite for knowledge, where does knowledge of the ontology come from? Turning and turning in a widening gyre ...

Why does Gruber cling to the word ontology, when he says that "it is certainly a different sense of the word than its use in philosophy", and he would in fact be better served by the term "epistemology" if he's concerned with knowledge? Because it sounds more down-to-earth, that's why. The earth is real, you see. On the published evidence here, Gruber is another Realist wrapped in the cloak of Formalism, trying to gate-crash the Groves of Episteme. "An ontological commitment is an agreement to use a vocabulary (i.e., ask queries and make assertions) in a way that is consistent (but not complete) with respect to the theory specified by an ontology." Tee-hee!

16 years on, Gruber has another definition, in which he writes that some people have said that "computational ontology [is] a kind of applied philosophy". Here, as often, philosophy seems to mean the thinking of deep thoughts, not familiarity with actual philosophers. Gruber does not appear to disagree with that view:

In the context of computer and information sciences, an ontology defines a set of representational primitives with which to model a domain of knowledge or discourse. The representational primitives are typically classes (or sets), attributes (or properties), and relationships (or relations among class members). The definitions of the representational primitives include information about their meaning and constraints on their logically consistent application. In the context of database systems, ontology can be viewed as a level of abstraction of data models, analogous to hierarchical and relational models, but intended for modeling knowledge about individuals, their attributes, and their relationships to other individuals.
Historical Background

The term "ontology" comes from the field of philosophy that is concerned with the study of being or existence. In philosophy, one can talk about an ontology as a theory of the nature of existence (e.g., Aristotle's ontology offers primitive categories, such as substance and quality, which were presumed to account for All That Is). In computer and information science, ontology is a technical term denoting an artifact that is designed for a purpose, which is to enable the modeling of knowledge about some domain, real or imagined.

The term had been adopted by early Artificial Intelligence (AI) researchers ... Some researchers, drawing inspiration from philosophical ontologies, viewed computational ontology as a kind of applied philosophy [10].

No comments: