How to publish datasets
You can proceed to publish the datasets, i.e. make them available to third parties. However, this step cannot be done by simply placing a link to the data anywhere on the website where it will be found online. A more appropriate place, especially for data published at a low level of openness, may be a specifically created Open Data section within your organisation’s web portal.
Cataloguing datasets has its own rules and when publishing datasets at higher levels of openness, it is necessary to create an appropriate distribution of the dataset to ensure that potential stakeholders are provided with the appropriate structure given by the chosen or created data schema. The publication of the dataset should be done in the following way:
- Determining the technical form of published data (called distributions),
- export of datasets in appropriate quality,
- Determination of conditions of use,
- export/edit data in open format,
- explanation of the structure of the exported data (metadata) and description of the distribution itself,
- the physical location of the dataset in a suitable repository (possibly in a local open data catalogue),
- creation of a cataloguing record in the local catalogue and/or NKOD.
Therefore, we recommend at a minimum to create a dedicated page that provides relevant information about the datasets and, in the case of publishing datasets available via APIs, the necessary technical documentation and technical details. The second option is to consider implementing your own local catalogue.
In case you are a small municipality or organization providing a small number of datasets (up to 10 datasets), we strongly recommend to choose the option to catalog datasets directly in the National Catalogue of Open Data (NKOD). Information on how to proceed can be found on the National Catalogue of Open Data website.
Obligation of cataloguing within the NKOD. However, it is not enough to announce the publication of open data only on your own website. According to Act No. 106/1999 Coll., it is obligatory to catalogue data in the National Catalogue of Open Data (NKOD).
Automating the preparation of datasets can also help significantly. In the case of automation, the IT department will again help. Typically, automation is handled by a script that creates a dataset of the required quality, saves it to a shared repository and alerts the curator to the existence of a new or updated distribution.
Local catalogue option
One way to create your own local catalog that will enable quality publication of your data files is to use one of the Open Data Management tools such as the opensource CKAN system. However, creating and especially running your own local catalog is technologically challenging and requires support from your organization’s IT department.
However, running a local catalogue has its advantages. Local catalog systems have fixed processes for data entry. They ensure standardization of open data publishing, quality assurance of metadata, availability of data sources, and adherence to recommended practices.
If your IT specialists have sufficient know-how, it is possible to operate a local open data catalogue that meets the requirements of the Open Data Catalogue Interface Open Formal Standard and register it with the NKOD. This allows the data provider to benefit from the possibility of automated updating of new records in the NKOD. This method of cataloguing saves the agenda associated with sending data messages to the NKOD for each dataset and also allows the presentation of the Provider’s data within its web presence.