Joel,
Your statement of
In some sections, Poli said that ontology deals with "objects", while epistemology deals with "concepts". What do you think of this?
Is not quite right. You may have misinterpreted section 1.4.1’s text (reproduced below).
In chapter 1 of the book of the Poli-Obrst reference [1] you cite, here are some excerpts:
From 1.2 Ontology_c:
A further point of contention or at least confusion is that between ontology and epistemology, i.e., on the study of what is vs. the study of what is ascertained and how it is ascertained. Ontology requires knowledge about what is, and if knowledge is described as, for example, justified belief, then ontology may be thought to devolve to knowledge and from thence to belief and justification for belief, i.e., the realm of evidence, manners and methods by which one adjudicates evidence to form belief, and thus epistemology. Ontology is not epistemology, but has a complex relationship to epistemology.
Ontology is primarily about the entities, relations, and properties of the world, the categories of things. Epistemology is about the perceived and belief-attributed entities, relations, and properties of the world, i.e., ways of knowing or ascertaining things. So epistemology is about empirical evidence gleaned that will be described or characterized by ontology.
1.3.2 Ontology_t and Epistemology
A further issue about ontology and epistemology should be brought out now, as
it relates to ontology_t. We have mentioned that epistemology deals with how
knowledge is known. How do my perception and understanding, my beliefs, constrain
my arrival at real knowledge or assumed belief, i.e., evidence, knowledge
hypotheses prior to their becoming theorems about knowledge (and there should be
a clear path from hypothesis to theorem to true theorem, but often there is not). So if
an ontology is a theory about the world, epistemology addresses the ways of acquiring
enough knowledge (and the nature of that) so that one can eventually frame
a theory. In ontology_t, the engineering artifact of the ontology model (a theory)
will require epistemological linkage to data. That data can be inaccurate, contain
uncertainties, and lead to partially duplicate but inconsistent instances of ontology
classes. Epistemology thus is employed in the use and qualification of data and as
stored in databases or tagged or indexed in documents.
If ontology states that human beings have exactly one birth date, the data
about a specific person is epistemological: in a given set of databases the person
instance named John Smith (we assume we can uniquely characterize this instance)
may have two or more attributed birth-dates, not one of which are known to be
true. Epistemological concerns distort and push off needed ontological distinctions.
Evidence, belief, and actual adjudication of true data is epistemological. What the
real objects, relations, and rules are of reality are ontological. Without ontology,
there is no firm basis for epistemology. Analysts of information often believe that
all is hypothesis and argumentation. They really don’t understand the ontological
part, i.e., that their knowledge is really based on firm stuff: a human being only
has one birth date and one death date, though the evidence for that is multivarious,
uncertain, and needs to be hypothesized about like the empirical, epistemological
notion it is.
In fact, much of so-called “dynamic knowledge” is not ontological in nature
(ontological is relatively static knowledge), but epistemological. What is an
instance that can be described by the ontology? How do I acquire and adjudicate
knowledge/evidence that will enable me to place what I know into the
ontological theory? Instances and their actual properties and property values at
any given time are dynamic and ephemeral (this particular event of speaking,
speaking_event_10034560067800043, just occurred; however the speaking_event
ontology class has not changed).
From 1.4.1 Developing Formalized Ontologies:
We may well read “ontologies” where Petrazycki writes “theories”. In fact, one
may well read “concepts”, since cognitive science has a comparable notion concerning
“concepts”, i.e., that they be non-profligate in a similar manner; especially
with respect to what is called the “theory–theory of concepts”, concepts as “mental
theories” (Laurence and Margolis, 1999, p. 43), and with respect to profligacy:
the potential concepts “the piece of paper I left on my desk last night”, “frog or
lamp”, “31st century invention” (Laurence and Margolis, 1999, p. 36). Interestingly,
it seems conceptual analysis is recapitulating ontology and semantics, since the former
is also addressing categorization, analyticity, reference determination, and the
notion of a “prototype” including the notion of “evidential” vs. “constitutive” properties
(Laurence and Margolis, 1999, p. 33), which stumbles on the epistemology
vs. ontology conundrum.
In chapter 2 [2] of the same book, I highlight a distinction:
2.1 Truth and Belief: Ontology, Epistemology, Contextual Semantics, Language, and Applications
Ontology is many things to many people, as the other chapters of these volumes demonstrate, and so no time is spent here defining ontology. This chapter focuses on architecture. One issue, however, needs to be raised: what ontology does not address particularly must still be addressed by ontology architecture. Ontology is not epistemology, nor is it the semantics of natural language. But aspects of these must be addressed by an account of ontology architecture. Why epistemology? Because, though ontology is about the real entities, relations, and properties of the world, epistemology is about the perceived and belief-attributed entities, relations, and properties of the world, empirical evidence gleaned that will be described or characterized by ontology. Why natural language semantics? Because, though ontology is about the real entities, relations, and properties of the world, natural language semantics is about the rendition in language of interpretations about the entities, relations, and properties of the world, and includes notions of sense and reference. In ontology architecture, epistemology is employed in the use and qualification of data and as stored in databases or tagged or indexed in documents. In ontology architecture, natural language semantics is employed in the analysis of natural language descriptions used to ascertain and represent the real world entities of ontology, the naming conventions used and the access to the interpretations about the real world that the ontology represents. One natural language processing technology in particular, information extraction, crucially depends on natural language semantics -- information extraction addressing the identification of entities, relations, and events in unstructured text, and the tagging or extraction of these to form instances of ontology concepts.
Of course epistemology is vastly simplified here, and is much more than that described above, since our main interest was (simply) distinguishing it from ontology.
Also note that John Sowa and many others had useful contributions to this book.
Thanks,
Leo
[1] Poli, Roberto; Leo Obrst. 2010. The Interplay Between Ontology as Categorial Analysis and Ontology as Technology. Chapter 1, pp. 1-26, in the book: TAO – Theory and Applications of Ontology: Computer Applications, Roberto Poli, Johanna Seibt, Achilles Kameas, eds. September, 2010, Springer.
[2] Obrst, Leo. 2010. Ontological Architectures. Chapter 2, pp. 27-66 in the book: TAO – Theory and Applications of Ontology: Computer Applications, Roberto Poli, Johanna Seibt, Achilles Kameas, eds. Springer. September, 2010, Springer.
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of joel luis carbonera
Sent: Wednesday, August 17, 2011 2:59 PM
To: ontolog-forum@xxxxxxxxxxxxxxxxSubject: [ontolog-forum] Artificial Intelligence, Ontology and Epistemology
In the area of artificial intelligence, we are interested in developing systems capable of exhibiting intelligent behavior. Some approaches to AI, especially the most related to the symbolic paradigm, assume that the knowledge about the world is a key aspect of the system to be able to do this. These approaches usually take a stance "representationalist" that there is an external world, a priori, to be represented internally using symbols. For these systems, the world is what can be represented symbolically in them.
Working in this context of AI, I raised several questions, which to me are intuitively related. But I still cannot verbalize this relationship explicitly. I'll list a few...
I always had doubts about the status of the relationship between ontology and epistemology in conceptual modeling and especially in AI. I found some articles, like those of Roberto Poli, that explore the relationship between these two areas:
"The Interplay Between Ontology as Categorial Analysis and Ontology as Technology"
In some sections, Poli said that ontology deals with "objects", while epistemology deals with "concepts". What do you think of this?
In foundational ontology, as the UFO, which deal with universals and particulars. Universals can be viewed as "objects" categorized by these ontologies?
Guarino also makes some statements about these issues, in his article: "Formal Ontology, Conceptual Analysis and Knowledge Representation". Then, I highlight an excerpt:
"Epistemology can be defined as “the field of philosophy which deals with the nature and sources of knowledge” [Nutter 1987]. The usual logistic interpretation is that knowledge consists of propositions, whose formal structure is the source of new knowledge. The inferential aspect seems to be essential to epistemology (at least for what concerns the sense that this term assumes in AI): the study of the “nature” of knowledge is limited to its superficial meaning (i.e., the form), since it is mainly motivated by the study of the inference process.
Ontology, on the other side, can be seen as the study of the organisation and the nature of the world independently of the form of our knowledge about it. Formal ontology has been recently defined as “the systematic, formal, axiomatic development of the logic of all forms and modes of being” [Cocchiarella 1991]. Although the genuine interpretation of the term "formal ontology" is still a matter of debate [Poli 1994], this definition is in our opinion particularly pregnant, as it takes into account both the meanings of the adjective "formal": on one side, this is synonymous of "rigorous", while on the other side it means "related to the forms of being". Therefore, what formal ontology is concerned in is not so much the bare existency of certain individuals, but rather the rigorous description of their forms. In practice, formal ontology can be intended as the theory of a priori distinctions:
• among the entities of the world (physical objects, events, regions, quantities of matter...);
• among the meta-level categories used to model the world (concepts, properties, qualities, states, roles, parts...)."
The Guarino's statement raises three issues on which I had been thinking. Two of them concern the relationship between ontologies, epistemology and AI, and one that is a philosophical question.
1-I am starting to work with ontologies. But by the definitions of ontology and the way they are presented in the literature, I suspected that the inferences are not subject of ontology. I'm working with expert systems in the field of Geology. In this expert system, from the description of the visual features of the rocks (described in terms of a domain ontology), one can generate interpretations about the physical processes that created this rock. This relationship between the visual characteristics of the rock and the physical processes seems a matter of epistemology. Am I correct? Can someone help me clarify this relationship?
2-How the ontology and epistemology are related in this case? Even though this relationship (between rocks and processes) is empirical, it uses the domain ontology. These inferences (occurring in the mind of an expert), seem to be a matter of epistemology. The inference seems to involve a visual comparison between the expert's knowledge (a pattern, perhaps represented in an inference rule) and what she/he sees. But ontology seems to play a structural role here. Without ontology, we would not have rock and process concepts. Am I correct? What do you think?
3-In the ontology, as philosophical activity, the ontology is an outcome of a top-down process, a bottom-up process or an interplay between both? It presupposes a relationship with the sensible world, or we are working with "a priori" contents? We perceive objects without first conceptualize them (turn them into a category in an ontology)? This question seems to be related to the philosophy of Kant. In ontology, what's more, my mind or the world? What do you think?
PS: My English is not very good, unhappily.