mintproject / model-catalog-python-api-client

MINT-ModelCatalogAPI-client
2 stars 1 forks source link

Extend the client to include DataCatalog API & Wings API #1

Open varunratnakar opened 5 years ago

varunratnakar commented 5 years ago

Create a wrapper around mint_client to return pointers to the DataCatalog and Wings APIs

brandomr commented 5 years ago

I believe that integration with the Data Catalog would be extremely useful.

Currently, to search for a model based on its associated datasets (using Data Catalog search functionality) you must:

  1. Find datasets in DCAT that match your query of interest (temporal, geospatial, standard variable) and grab the dataset IDs
  2. Find all model configurations from MCAT
  3. Subset model configurations to only those that have input/output datasets matching the dataset IDs found in #1
  4. Find all model versions from MCAT
  5. Subset model versions to only those that have configurations found in #3
  6. Find all models from MCAT
  7. Subset models to only those that have versions found in #5

This process assumes that dataset IDs are consistent between DCAT and the Model Catalog. This can be done through middleware between these two systems, but should be managed by one or the other (either DCAT or the Model Catalog).

This process should be streamlined so that search for models (based on associated input files or output files, as queried through DCAT) can be executed via this API client. cc @dgarijo @jgawrilo @dnfeldman