[29]                             home                             [31]

 

Thursday, March 16, 2006

 

Challenge problem à

The Taos Discussion à

Key-less Hash table à

The Second School of Semantic Science

 

ontologyMapping Glass Bead Games

 

Communication from Azamat Abdoullaev

 

 

Paul,

 

Comments from Paul on regularization and BCM à [31]

 

As you know, the distinction between semantic networks and ontologies implies a more general difference between knowledge representation technologies and languages (as FOL, semantics nets, frames, rules, DL, SW languages) and foundation Ontology as the general science of entities and relationships.

 

In computer science, artificial intelligence, and software engineering, the science of reality [studying how to represent and reason about the world, both external and conceptual] is commonly (or confusedly) regarded as an extension or an external layer of logical calculi and formal languages. As a consequence, a real world ontology is reduced to a sort of formal logical ontology, which is specified as consisting of the following logical elements: concepts (classes, objects, or categories) with their characteristics (attributes, slots, functions, roles, or properties) and relations (generalization and specialization, functions) restrained by logical axioms (assertions) and exemplified by instances of classes and specific properties.  [1]

 

Without the reality check restraints, our intuitive, individual ontologies may be as various as: 'an explicit specification of conceptualization', 'a theory of content', 'a theory (a system) of concepts/vocabulary used as building blocks of information processing systems', 'a set of agreements about a set of concepts', or 'the representation of the semantics of terms and their relationships'. Or, 'the class hierarchy in object-oriented paradigm', 'a complete schema of the domain concepts', 'an entity-relationship schema with subsumption relations between concepts'. You may risk meeting also such definitions as 'conceptual patterns', 'concept heterarchies or hierarchies', 'a body of conceptualizations', 'schemata', or 'metadata scheme', 'a common set of terms', 'a controlled vocabulary of terms', 'a representation vocabulary', or 'a body of knowledge'.

 

Such a transient state of affairs may be partly justified by the long-standing disagreements over the scope and nature of the subject matter even among its greatest scholars.

 

For, when taken as pure and abstract knowledge, Ontology is formulated as different as:

 

·        the science (account) of entity (or being) in general;

·        the knowledge of the most general structures of reality;

·        the theory of the kinds and structures of things in every domain of reality;

·        the study of entity types and relations;

·        the most general theory concerning reality, being, or existence;

·        a collection of absolute assumptions; the study of change;

·        the science of all possible worlds and everything conceivable;

·        the study of semantic values of natural and formal languages and

·        ontological commitments about the world.

 

This inherent ambiguity causes the researcher to decide whether the whole activity is about the inquiry of entity, its forms and properties, or just about some general concepts with their formal logical relations.

 

Or, are we supposed to deal with 'the nominal' ontology of terms and their semantic relationships instead of 'the world' ontology of entity types and their external relationships? Because of this manifold interpretations, one has to decide on the research line: either the world ontology (realistic and veridical), or the concept ontology (conceptual and notional), or the word (terminological) ontology (linguistic and nominal).

 

In fact, there is a unified foundation ontology (UFO) dealing with the world, its things, beings, and relationships, and a plurality of domain ontologies dealing with the specific regions, parts, domains, or realms of reality.

 

Accordingly, computing ontology is all about the representation of the world, its entity states, changes, and relationships, in machine-processed forms. In other words, there is Ontology taken in the primary sense and ontologies in the secondary senses. Its basic meaning consists in being an account of reality and realities and their associative orderings. Thus, Ontology is primarily concerned with the types of entities and relationships in the world. Secondly, it studies how things in the real world relate to concepts and associations in the mind, to coded representations and structures in machines, and to words and sentences in natural languages.

 

At the end, I’d like to refer all of us to a seminal work of Randall Davis & co-authors, What is a Knowledge Representation? AI Magazine, 1993, which is generally in line with the positive view, except maybe one traditional for logicians confusion that ''the "world" we are interested in capturing is the world inside the mind of some intelligent human observer (e.g., a physician, engineer, etc.)''. Here are some characteristic selections:

 

''a KR is a set of ontological commitments'', the intuitive assumptions about how to view and reason about the world;

 

''Ontologies can of course be written down in a wide variety of languages and notations (e.g., logic, LISP, etc.); the essential information is not the form of that language but the content, i.e., the set of concepts offered as a way of thinking about the world. Simply put, the important part is notions like connections and components, not whether we choose to write them as predicates or LISP constructs.'';

 

''... all the representation technologies...supply only a first order guess about how to see the world: they offer a way of seeing but don't indicate how to instantiate that view. As frames suggest prototypes and taxonomies but do not tell us which things to select as prototypes, rules suggest thinking in terms of plausible inferences, but don't tell us which plausible inferences to attend to. Similarly logic tells us to view the world in terms of individuals and relations, but does not specify which individuals and relations to use''.

 

Kind regards,

 

Azamat Abdoullaev

 

EIS Encyclopedic Intelligent Systems LTD

 

 

 

 

 

 


[1] The logical elements that you list are complete, except sometimes one talks about facets which are a typing of slots.  All of the names of logical elements have variation in meaning as one moves from one researcher to another.  I strongly claim that the logical elements themselves are ontological assertions, or come with implicit ontological assertions (which vary from researcher to research).  The ontological assertion that a parent – child relationship is “ontological”, i.e. a pure reflection of THE reality, is but the most obvious assertion of RDF based systems.  The use of this assertion becomes a high art form in the Protégé community and other knowledge engineering communities.  There is clearly value in organizing data and information in a hierarchical fashion using subsumption.  But if one forgets that the subsumption itself is an assertion made by the ontology designer; it is natural that the implicit information that is derived by deductive inference will be shaped by that subsumption assertion.