negation of relation names

In the relations ontology we express:
eukaryotes consists of nucleus  ==> consists_of(eukaryotes, nucleus)
eurkaryotic cell enveloped by cell membrane ==> enveloped_by(cell, cell membrane)

as I am representing the school textbook, I find a few sentences to be of the form of negation. These are:

prokaryotes does not consist of nucleus ==> ~consists_of(prokaryotes, nucleus)
prokaryotes not enveloped by membrane  ==> ~enveloped_by(prokaryotes, membrane)

Such negations are useful for students to understand the difference between eukaryotes and prokaryotes and cannot be avoided during representation.

so my query is: can we include the negations of relation names in the relations ontology. how can we represent such negation relation names.

since ~consists_of, ~enveloped_by are negation of consists_of, enveloped_by respectively, does the relation names ~consists_of, ~enveloped_by still fall under the meronymic inclusion and spatial inclusion  respectively.

I think so.  In the above assertions, we know in reality that prokaryotes donot consists of nucleus, but this is equally important for representation. In cases such as something may cease to exist at a particular time t then we still have to represent such assertions even though these would be in negation forms.

But the question that is grappling my mind is the following:

ontology is the representation of types. it represents what exists in reality. now something that does not exist in reality, and yet if we want to represent it then would we still be creating ontology of non-existent entities. and if yes, then would it still be  ontology.

Perhaps, philosophers can  provide a convincing argument for my question.

relations and functions in biology

As we know that functions (1 to 1) are different from relations (1 to many). All functions are relations, but not all relations are functions. In the case of creating an ontology for biology, we use relations. But, in biology, is it possible to apply the functions as well. Can there be any such relation which is a function in the case of biomedical ontology. For example, part-of, composed-of, located-in, surrounded-by, etc. are all relations in the context of biomedical ontology. Or do we have to always understand biology ontology in terms of relations only and not functions.

Prof. Barry Smith (University of Buffalo) clarifies the distinction between ontology of relations and the way relations and functions are treated in a set theory. Further he emphasizes on the functional associations (in mathematical sense) can occur in process relations. For example, process-a regulates process-b.

Now this sounds interesting. In this case, I think, some of the process relations can be functional. Such process relations are : regulates, transforms into, derives from, develops into, preceded by, results into, etc.

Let me illustrate with a few assertions:
Larva transforms into pupa
zygote develops into foetus
blastula transforms into gastrula

What I am trying to point out is that the domain and range for these process relations will always be of  1 to 1 mapping. So, these process relations can be of functional type.

But the structural relations such as part of, surrounded by can not be of functional type as a domain can have many ranges in such relations.

P.S. This post is based on my discussion on relations and functions in biology on the OBO mailing list.

what is so different about GNOWSYS

Gnowsys can be used to represent structural knowledge, process modelling, as far as scientific knowledge is concerned. But then there also exists several other modelling tools such as—OBO-edit, Protege, OPM, KEGG, Virtual Cel, Cell Designer, Reactome, GO, etc. which can as well represent knowledge. So then, an obvious query that comes to my mind is  what is so different about Gnowsys? Isn’t it just an addition to the existing modelling tools. Nagarjuna explained that the differences are in the form mentioned below:

One of them is the architecture of Gnowsys itself being different with the other tools. As per the gnowsys architecture, everything that goes in is a node for e.g. concept, concept type, instance, relation, relation type, attributes, attribute type, are all stored as nodes, but in the form of networked nodes. A node cannot have meaning on its own. Its meaning is established only by its neighborhood. For instance, X is related to Y, and it can have many such simple relations.  This is in accordance, with the theory that knowledge exists in a network. If we want to understand what is force, then its understanding can be acquired depending on what relations does the concept force shares with other concepts like, mass, inertia, acceleration, etc.

Gnowsys can create versions even if there is a single (simple) change to a node. This change can be in the form of either adding a node, editing a node, adding a relation, editing a relation, deleting a relation. From this list even if there is change in just one kind, then also it can create versions and record the history. This kind of change is accordance with the degrees of conceptual change that Paul Thagard explains in his book Conceptual Revolutions. According to Thagard,  conceptual change can occure when there is addition of concepts, addition of relations. With this model of versioning, gnowsys can record the changes that ocurs in people’s conceptual framework, thus enabling to trace the conceptual changes.

Another difference is that gnowsys does not store knowledge in any kind of file system. It is on the web, in another way to put it, gnowsys does not ask in which file format should it save the knowledge created. But at the same time, it can possibly exchange to various formats that of OWL.

The other difference is that gnowsys can store multiple ontologies. Using just one application of Gnowsys, it would be possible to store structural knowledge, process, metabolic pathways, etc. It will be interesting to see the multiple ontologies by importing a few of the existing ontologies (can be cell division, gene ontology, metabolic pathways, environment ontology,  etc.)  from the public domain and from several different places and allow these ontologies to store in gnowsys.

It would be useful to know, if there are other differences.

This excerpt is based on discussion over tea.