Collections Best Practices

[rebeccawilliams]

As requested at the Interagency Open Data Meeting: What are the best ways to organize data into Collections for users?

Introductory guidance is here: https://project-open-data.cio.gov/v1.1/collections/

Things to consider:

• user testing results
• temporal, spatial considerations
• existing hierarchies and schemes (including Bureau and Program Code)
• emerging data.json practices
• supportive tools, like Inventory.data.gov
• ways Collections can address duplication, if any

[rebeccawilliams]

Note: Currently Data.gov does not support collections via geospatial harvest sources without registering an individual WAF harvest sources for the collection.

Data.gov is working to fix this issue and you can follow along at: GSA/data.gov#708, but for now the only option is manually registering collections via a separate WAF source (or via Project Open Data Metadata). The fix to support collections in geospatial metadata will focus on ISO 19115 via a CSW. Support for collections from CSDGM metadata is more uncertain since no mapping for the field has been defined.

Long-term, shared services w CKAN/Geonode integration may be able to streamline reconciling the various geospatial/project open data metadata crosswalk issues.

[rebeccawilliams]

There is guidance on how to indicate a collection (parent/child) with the Inventory.data.gov tool: http://www.digitalgov.gov/resources/inventory-data-gov-guide/

The last two fields are to indicate if this dataset is part of a collection or if you want to designate this dataset as the “parent” of a collection.

Is parent. Yes or No. If you want to make this dataset the “parent” of a collection, choose yes. Otherwise choose no.

Note that once you make a dataset a “parent” dataset, it cannot be a “child” dataset, or part of another collection of datasets. After a dataset is selected as parent, this dataset will be available for the subsequent datasets in this organization to choose a parent dataset (for isPartOf)

If you chose No and made the dataset a non-parent dataset, you would then be able to choose in the next field (Parent dataset) – the datasets that you want to make this dataset as part of a collection, note that this field is optional and can be left empty if you would like the dataset to be an independent dataset.`

[rebeccawilliams]

All of the above guidance has been incorporated into #547.

[rebeccawilliams]

Reopening this until #547 is merged.