Data Dictionaries for Humans and Machines

Shared datasets often have column/field names that are ambiguous in their meaning, or contain identical/related concepts with different names, hindering reuse. This ambiguity happens regardless of the method of sharing – via files, web pages, or APIs. The traditional solution for this is to provide documentation.

However, documentation is typically delivered “out-of-band”, i.e., not directly linked within the same artifact (file/object) as the data. Furthermore, any linking among terms by different data providers, whether to assert equivalence or to clarify differences, is typically done in an ad hoc manner, hindering interoperability and thus reuse.

I am getting started with a working group of materials scientists to develop data dictionaries using Linked Data principles. I’d like to highlight one feature of the HTTP standards that is helpful in serving both human-readable and machine-readable documentation for data:

In HTTP, content negotiation is the mechanism that is used for serving different representations of a resource at the same URI, so that the user agent can specify which is best suited for the user (for example, which language of a document, which image format, or which content encoding).
MDN Web Docs

For example, let’s say we have a data dictionary identified as with a contained term When you click on the link in a web browser, your user agent (e.g. Firefox, Chrome, Safari, etc.) assumes you want a response in the “text/html” format. However, if you are accessing the resource programmatically, you may want a different format:

# A readable, hand-editable RDF format ("turse [sic] triples")
curl -H "Accept: text/turtle" \
curl -H "Accept: application/ld+json" \
curl -H "Accept: application/rdf+xml" \
# Or explicitly request HTML
curl -H "Accept: text/html" \

Not all web servers implement content negotiation. However, it is a nice alternative to making data consumers scrape web pages, find links by eye to obtain different formats, or remember your convention for how to alter a URL to obtain the desired resource – it’s the same resource, just a different representation.

You can have a look at the implementation of the above “hello world” example here. Thank you Matthew Evans for contributing to the code.

Subscribe to get short notes like this on Machine-Centric Science delivered to your email.