When talking about open data the 5-star principle has paved our way. And although the highest level, linked data, is machine readable, it is not application friendly from a reusability perspective. By "reusable applications" I mean applications (mobile apps, server apps, web-apps, etc) that work on multiple data sources from different data publishers without configuration changes.

For example, an application to visualize results of local elections would be far more reusable if there were a data format for election results.

I am looking for data standards for open data that applications can use to operate on multiple data sources. I could not find any standards directories nor standardizing bodies that drive this process.

This could be a killer feature for the open data movement since, on one hand, application developers could rely on a single format description, and on the other hand data publishers get applications for published data faster and cheaper.

Do we already have such "domain" standards for open data formats? Am I missing something?

2 Answers 2


Your question is quite open ended and answers may be quite subjective, but here are some resources and pointers.

There are in fact many different domain specific standard data models and exchange formats. You can find some of these in standards registries like fairsharing, which may have a bias towards scientific and clinical standards, where there may be more incentives towards producing standards.

I can imagine producing a single standard for election results might be hard. For some applications, some aspects such as location may be captured in existing standards like GEOJson. It may be hard to include multiple different countries voting schemes and rules in one standard.

One strategy here is to standardize at the level of metadata elements or vocabulary terms. The ISO 11179 seeks to do this, and linked data standards also seek to make it easier to make data self describing, with data described in RDF or JSON-LD, using one or more standard vocabularies. However, after two decades, uptake of these approaches is patchy, and variable across communities, for reasons that are complex and widely debated.

Newer general purpose standards like frictionless, LinkML (disclaimer: I am involved with this effort) seek to address these but having a one size fits all answer is hard, even agreeing on the base substrate is impossible. For some communities CSV plus standard data dictionary elements may be the answer, for others, dedicated JSON schemas or SQL databases, others with high volume data may need HDF5 based substrates.


There are many different domain-specific standard for open data. There are also general metadata frameworks which maybe are what you are thinking about: see for example frictionless which provides a framework & and a standard for writing metadata (in json format) describing tabular data (e.g. csv files).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.