Vision: only computational models?

[platipodium]

Do we only "need a new generation of scientific computational models" or do we need also conceptual models? please Discuss here!

[platipodium]

Followed up in #345

[hrajagers]

I think that there is a two-way interaction between computational models and conceptual models.

1. The standards help to abstract away from the details of the computational models, bringing them more on par with conceptual models. For instance: this component works on this kind of domain, with these input variables, and these output variables. Those ideas are also applicable for the conceptual models.
2. If we cannot come to a coherent well defined conceptual model for what we want to achieve, we will not be able to develop a proper technical model.

However, that does not mean that every technical standard for computational models is equally applicable to conceptual models. The computational implementation will probably need far more standards than the conceptual models.

[BruceEdmonds]

It is hard to standardize conceptual models, unless they adhere to a particular formalisations - they are so various. On the other hand including all the other models a model relates to should be mentioned, but making this is different from them being first-order entities we are focussed upon

[chporter]

The conceptual model might best be described as part of the documentation of a computational model.

[platipodium]

So that would strictly be the textual representation of a computational model? I think there's more to the conceptual modeling term. I could think of

1. A precursor stage for a formalised model
2. A model that also includes beyond the formalisation the goal of the modeling

[BruceEdmonds]

there are also sub-models, meta-models, reproductions of models etc. etc.

[aavoinov]

there are also sub-models, meta-models, reproductions of models etc. etc.

Sure, all sorts of models out there, but does it mean they need special standards? Same standards can easily apply to a sub-model. However conceptual models are quite a different type of a beast that deserves some attention. IMHO.

[percivall]

Each computational model should have a conceptual model. Development of an OMF Framework can develop by comparing conceptual models from the several computational models. Where OMF members agree, OMF Conceptual Models for various topics can emerge. For example, having OMF conceptual model agreement on spatial referencing enables interoperability between models.

[BruceEdmonds]

Definitely agree that each computational model should have a conceptual model - but this might be composed of sub-models etc. of many different kinds - some mathematical, some other computational models, some theoretical, some linguistic....

[MarcoAJanssen]

Conceptual models are indeed important, but I think that we need to have boundaries for an organization, and a vision statement focusing on computational models is doing this. Other types of models, conceptual, analytical, etc, can be used to document and formulate our computational models. I like to keep the vision focused on computational models.

[aavoinov]

Oh, so analytical models are also out? And if I run an analytical model using a numerical method, then they are OK for us? I'm lost. I'd rather start broad instead of excluding certain people and groups upfront.

[MarcoAJanssen]

no they are not out, we do not emphasize them in the vision statement.