(I chose the Wikipedia in both cases as giving a wide, "bottom line" account)

Here, i want to talk the main idea of the ontology in computer science area. Recently, Ontologies have been developed to provide a machine-processable semantics of information sources that can be communicated between different agents (solfware and humans). The idea of ontologies is to allow users to organize information on the taxonomies of concepts, with their own attributes, to describe relationships between the concepts. When data is represented by using the ontologies, software agents can better understand the content of the data and messages, smarter integrate data for multiple jobs than the conventional web used. In this way, the sites not only performs the function of receiving and displaying information but also the ability to automatically extract information, query, argue the knowledge base to give accurate information automatically.

When used in knowledge representation, ‘ontology’ is largely a matter of mapping conceptual domains. Computer science ontologies are epistemic in scope - from the philosophical point of view, they're concerned more with 'sense' than with 'reference'. More generally, upper ontologies and domain ontologies allow us to model and manipulate the relations between concepts: that is, they allow us to determine nodes in epistemic and semantic *webs* or *networks* ('top-level' and 'local' holisms). However, they're limited by their metaphysical assumptions (this is notable in the distinctions between 'continuants' and 'occurrents’ and in the naïve assimilation of ‘continuants’ with ‘endurants’ and of ‘occurrents’ with ‘perdurants'). I also suspect that there's a serious confusion between "use" and "mention", and particularly among CS ontologists who lack philosophical training.

Given this underlying referential inflexibility, even the most adaptable ‘applied ontology’ will have weaknesses. We can establish ontologies that allow users to personalise the determination of epistemic and semantic nodes, which is all well and fine if our sole objective is to provide more efficient semantic and epistemic tools for everyday applications (everything from the semantic web to expert systems). However, such tools are simply a means of adapting the system to existing habits or patterns of thought - they can’t help us develop innovative or alternative ways of thinking about the world.

Without going into the highly technical details of the philosophical debate, metaphysical ontologies are more generally concerned with ‘entities’ rather than with ‘concepts’; in my personal view, I’d hold that a metaphysical ontology is rather concerned with ‘particular individuals’ than with ‘instantiated universals’. Thus, while a CS ontology is concerned with concepts, a metaphysical ontology is concerned with the (putative) *extension* of concepts. This reproduces the sense/reference distinction I mentioned above: CS ontologies are concerned with sense within language, while metaphysical ontologies examine the mechanisms of how we project our language onto the world.

Metaphysical ontologies project linguistic terms onto ‘the world’ either by naming things that are in some way “self-evidently” individuals or by constructing individuals in space and time. The latter alternative allows us to project individuals by geometric and topological analogies: an individual I is a construct on some space S (that is, S is the *domain* of which I is an element). The ontology of such ‘projected individuals’ is largely concerned with their extension, their composition, and their persistence; for this reason, mereological analysis has become a central element of contemporary metaphysical ontology - though, as the ontology concerns both the boundaries and the part-of relations which determine and organise entities, we should perhaps more properly speak of ‘mereotopological’ analysis (cf. http://en.wikipedia.org/wiki/Mereotopology and, for a fuller development, Barry Smith’s paper http://ontology.buffalo.edu/smith/articles/Mereotopology1.pdf). Such approaches project individuals onto some region R of a domain S such that the boundary between that region and its local environment depends on a topological relation between R and its parts. Following the geometrical analogy, S can be a classical Euclidian space, a temporal interval, a 4D spacetime, or a phase space (or any other geometrical space, in the ontologies of physical science). Projections onto classical space, onto time, onto spacetime, and onto phase spaces are of great interest in constructing the ‘ontologies of everyday life’ required by applied philosophy (and by systemics in particular).

Mereotopological approaches are available in computer science ontology; nevertheless, such ontologies generally remain ‘descriptive’– that is, they attempt to account for pre-theoretical or unanalysed intuitions and assumptions about supposed “matters of fact about the world”. This is particularly the case when certain assumptions are built into the model – when, for example, a ‘human individual’ is given as a continuant, and continuants are subsumed to endurants. However, mereotopological analysis not only allows us to describe such intuitions and assumptions - it also allows us to construct alternative models of how some individual might be projected onto a given domain (to give a simple example, certain systems can be modelled both as occurrent and as continuant individuals). Insofar as systemics is concerned, such "constructive ontologies" allow us to model systems with respect to their 'adjacent possibles', thus permitting a projection of how the system might evolve. I have no clear idea what implications this might have for cybernetics, but in terms of organisational systemics it's extremely promising, and particularly in the field of change management and the dynamics and structure of professional organisations.

(This last remark is a general repositioning of my initial remarks with respect to my latest research; I can of course develop any points and - I hope - answer quite a few of the numerous objections it raises)