Christoph Helma wrote:
> Excerpts from Yap Chun Wei's message of Wed Oct 07 10:03:28 +0200 2009:
>   
>> I have just joined a few days ago so pardon me if I am saying something that
>> has been discussed before. Would it be better to adopt the standards used in
>> PMML for representing datasets? For example, if you look at the specifications
>> for the SupportVectorMachineModel, we can use the same idea of how it stores
>> the support vectors using VectorDictionary to store a dataset. Just have to
>> couple it with the DataDictionary so that information about the features are
>> available.
>>     
>
> Yes we have discussed that before - PMML is not (yet) generic enough to
> cover all of our models. One possible option was to create PMML
>   
Plus, there is no support in PMML for including URI to dataset and other
types of resources, though adding such could be considered another
extension.

There is a general consensus within OpenTox (correct me if I am wrong)
that PMML support is important, but mostly for exporting models, rather
than internal exchange format.

Best regards,
Nina
> extensions, but as far as I can see it, we did not come to a defintive
> conclusion yet.
>
> Best regards,
> Christoph
> _______________________________________________
> Development mailing list
> Development at opentox.org
> http://www.opentox.org/mailman/listinfo/development
>