Snapshot of the BioZEN ontology
Matthias Samwald
Hello,
I have attached a snapshot of the 'BioZEN' ontology that I have been playing around with in the last month. It is based on a subset of the DOLCE top-level ontology and a subset of SKOS. It was derived from these two foundational ontologies and BioPAX level 2. I merged all of the ontologies and gradually mapped BioPAX so it was in accordance with the basic DOLCE framework (inevitably, I also added a lot of my own ideas, of course).
There are still many flaws in this ontology and some things have yet to be implemented, but it should give you an idea of where it is heading.
Two major design principles:
* The ontology is based on OWL individuals and spatiotemporal particulars, NOT on classes. The end-user should not need to define OWL classes. This changes the semantics of this ontology quite fundamentally when compared to BioPAX. The users are supposed to describe biological systems through concrete EXAMPLES ('model systems'), not by making general statements. If the end user wants to make broader statements, he does so by describing canonical model systems (that are described just like concrete spatio-temporal particulars nonetheless). This makes working with the ontology much more intuitive and less complex - especially for users that do not know too much about ontological design principles. It also makes it possible to extend the ontology to describe simulations of biological models (similar to SBML or CellML).
* The ontology distinguishes between 'things' and 'concepts of things' - the latter being described through concept hierarchies expressed in the SKOS core ontology. The hierarchies of concepts are used to annotate the 'things' we are describing.
This has two advantages: First, it takes away the necessity to define complex OWL class hierarchies. All the cells we want to describe are simply instances of the class 'cell'. If we want to provide additional information about the type of cell, we are annotating the instances with cell-type-concepts (e.g. with the concept 'Purkinje cell', which is a narrower concept of 'Neuronal cell'). This helps keeping the class structure lean and clean and offers users the ability to create such hierarchies without defining new classes. Second, this approach seeks to unify the advantages of stringent ontologies (consistency, better processing by reasoners and software) with the advantages of taxonomies and 'folksonomies' (easy to use, flexible, less side-effects).
Notes:
* The ontology imports OWL files from the internet so you have to be on-line to load it.
* For some strange reason, Protege does not show the right list of properties for each class, i.e. it also shows properties that do not have the class as its domain. Therefore, the list of allowed properties for each class appears longer than it actually is.
//Matthias Samwald
Alan Ruttenberg
Thanks for this. It's a lot to absorb, and I wonder if you have some
examples of individuals you have modeled with it, and what inferences
are made that you find useful. It might be worth scheduling some time
to teleconference and have you talk to us about this.
Regards,
Alan
Matthias Samwald
kind regards,
Matthias
> --~--~---------~--~----~------------~-------~--~----~
> You received this message because you are subscribed to the Google
> Groups "BioPAX-Manchester" group. To post to this group, send email
> to BioPAX-M...@googlegroups.com
> To unsubscribe from this group, send email to BioPAX-Manchester-
> unsub...@googlegroups.com For more options, visit this group at
> http://groups.google.com/group/BioPAX-Manchester -~----------~----~-
> ---~----~------~----~------~--~---