[OTDev] Representing parameters with optional values ...

chung chvng at mail.ntua.gr
Sun May 15 16:18:55 CEST 2011


On Sun, 2011-05-15 at 01:43 +0530, surajit ray wrote:

> Hi pantelis,
> 
> 
> 
> On 15 May 2011 01:15, chung <chvng at mail.ntua.gr> wrote:
> > Hi surajit,
> >    First of all I think it's impossible to have the XSD datatype there
> > except if you declare it additionally. We can have a property that maps
> > ot:Parameter-type individuals to XSD datatype URIs or we can subclass
> > ot:Parameter and introduce ot:StringParameter etc. Instead, I would ask
> > what is the purpose of having the datatype in there. And the answer is,
> > to provide information to the client regarding the expected type for the
> > particular parameter. This however can be retrieved by an OPTIONS on the
> > resource  (see latest ontology update and specification for OPTIONS).
> > We've already implemented this method for most of our resources.
> > RDF-formatted OPTIONS can even lead to automatically generated web
> > interfaces for web services!
> >
> 
> Could not find any page on OPTIONS on opentox.org (I searched with
> "options") . Could you point me to the right page ...


We haven't put that online yet. It was just some email we had exchanged
on the mailing list. You can take a look at our implementation ( we've
updated ToxOtis so you can use it or check the source code -
https://github.com/alphaville/ToxOtis ). I have updated the OpenTox
ontology in collaborative Protege so you can take a look therein too.

> 
> > Evaluated runtime is not a parameter in the OT sense as it is not
> > something expected by the client as input. This can be assigned to your
> > node using some other ontological property (e.g. ot:evaluatedRuntime) or
> > you can include it only in HTML representations and not in RDF.
> 
> There are some parameters in SVM model making that can have a default
> value based on the number of compounds or some other factor. However
> the underlying software can take a value for that parameter. In that
> case the evaluated value for the parameter is not used.
> 
> >
> > Best regards,
> > Pantelis
> >
> > On Sat, 2011-05-14 at 11:43 +0530, surajit ray wrote:
> >
> >> Hi Nina, All,
> >>
> >> For the algorithm API we have to attach parameters which are
> >> represented with a scope, a default value (with type) , and the name
> >> and description.
> >>
> >> However some parameters may not have a default value. For example if
> >> the default value is a function of another set of parameters OR it is
> >> evaluated runtime (although may be overriden by a fixed value by the
> >> user). In such cases there will be no default value. In this case how
> >> to we mention the XSD TYPE. (See the comented line below).
> >>
> >>
> >>             Individual parameter31 =
> >> jenaModel.createIndividual(OT.OTClass.Parameter.getOntClass(jenaModel));
> >>             //parameter31.addLiteral(OT.paramValue,
> >> jenaModel.createTypedLiteral("", XSDDatatype.XSDfloat));
> >>             parameter31.addLiteral(OT.paramScope,
> >> jenaModel.createTypedLiteral("optional", XSDDatatype.XSDstring));
> >>             parameter31.addLiteral(DC.title,
> >> jenaModel.createTypedLiteral("integer_float_cutoff",
> >> XSDDatatype.XSDstring));
> >>             parameter31.addLiteral(DC.description,
> >> jenaModel.createTypedLiteral("If Integer or Float Feature what is the
> >> cutoff ? Float value required ", XSDDatatype.XSDstring));
> >>             MCSSFinder.addProperty(OT.parameters, parameter31);
> >>
> >>
> >> To put it simply ...Since the type is associated in the declaration of
> >> the default value, what is the mechanism to mention the type when
> >> there is no default value ?
> >>
> >> Thanx
> >> Surajit
> >> _______________________________________________
> >> Development mailing list
> >> Development at opentox.org
> >> http://www.opentox.org/mailman/listinfo/development
> >>
> >
> >
> > _______________________________________________
> > Development mailing list
> > Development at opentox.org
> > http://www.opentox.org/mailman/listinfo/development
> >
> 
> 
> Also I have a question about Model API. Should the ot:hasSource on it
> should point to the task object that generated it ?
> [The ot:hasSource for the task contains the algorithm that created it]
> 
> Regards
> Surajit Ray
> _______________________________________________
> Development mailing list
> Development at opentox.org
> http://www.opentox.org/mailman/listinfo/development
> 





More information about the Development mailing list