On 31 March 2011 13:48, Christoph Helma <helma at in-silico.ch> wrote:

>
> > > Thanks, this works as advertised! But how can I decide, which datasets
> > > are ready for production use (e.g. to select the four relevant datasets
> > >
> > >    dc:title "Benchmark Data Set for in Silico Prediction of Ames
> > > Mutagenicity" ;
> > >    dc:title "Bursi mutagenicity dataset.sdf" ;
> > >    dc:title "CPDBAS: Carcinogenic Potency Database Summary Tables - All
> > > Species" ;
> > >    dc:title "ISSCAN: Istituto Superiore di Sanita, CHEMICAL
> CARCINOGENS:
> > > STRUCTURES AND EXPERIMENTAL DATA" ;
> > >
> > > from 247 mutagenicity datasets), if I do not know titles in advance?
> > >
> >
> > There is currently no any metadata to handle it,  let's agree on some RDF
> > property  to denote a "production use" and we'll include it it into
> > /dataset/id/metadata ( read and update) , as recently was done for
> licenses.
>
> That would be *very* useful, also for models (or even algorithms - to
> distinguish between algorithms in development and stable/mature
> algorithms).


Agree.  Could be applicable even for chemical structures.


> Do you know existing properties that can be used or do we
> need to invent one?
>

Not really, now trying to look into versioning / provenance ontologies like
this one http://www.w3.org/2005/Incubator/prov/XGR-prov-20101214/  , but not
sure.

If anybody else on the list has a better idea, please do send a pointer.

Best regards,
Nina


>
> Best regards,
> Christoph
>