Data Publishing

Since harvests datasets from national portals, it is not used directly by civil servants to publish data. Rather, an entire portal, such as the Romanian portal, is added to as a 'harvest source', after which it will automatically be regularly polled for updates. If the portal being added is CKAN-based, then the process of setting up harvesting for it takes only a few minutes. However, it is worth briefly explaining the process of publishing data on a source portals. Government employees have individual accounts with authorisations depending on their department. A web user interface guides an authorised user through the process of publishing data, either uploading or linking to the data itself and adding metadata such as title, description, keywords, licence, etc., which enable users to find and identify it. A published dataset may have a number of data files and displays the publishing department as well as the other added metadata. For most common data formats (including CSV, XLS, PDF, GeoJSON and others), users can not only see the metadata but also preview the data in tables, maps and charts before deciding to download it. The process described is for using a CKAN portal, but other data portals will have similar functionality.

Importantly, publishers are not constrained to publish only linked data, or only data in a particular format. While publication of linked data may be desirable, it is far more desirable for data to be published than to be not published. Pragmatically, this often means governments making data available in whatever form they have. One of the ways in which adds value to published data is with a feature (described in more detail in Sect. 3) to lift data in the simple CSV spreadsheet format into linked data's native RDF. Two projects by governments to publish well-modelled, high-quality linked data are also described later in this chapter.

Data Consumption

Users on can search for data from any of the harvested portals[1]. A search is made by specifying particular words or terms. CKAN searches for matching terms anywhere in a metadata record, including for example the description, and returns results (datasets) ordered by relevance or date modified. The user can add filters to the search to find only datasets from particular countries, or with particular tags or available file formats. A top-level set of nine topic areas (Finance & Budgeting, Transport, Environment, etc.) is also displayed and can be used for filtering search results. As with the source portals described above, they can preview the data in various ways before deciding to download it. Users can search, preview and download without registering, but registering enables a user to follow particular datasets or topics, being notified when the dataset changes or when new datasets are published.

The user interface of CKAN has been translated into a wide range of languages, and users can choose the language in which they interact with the site. Like all CKAN instances, can be accessed via an RPC-style API[2], as well as via the web interface. Metadata for each dataset is also available as linked data, in N3 or RDF-XML format[3]. It is also possible to get a dump of the entire catalogue as linked data. As mentioned above, the site also includes a feature for lifting data published in CSV spreadsheets to RDF.

  • [1] As of June 2014, there was 1 pan-European, 13 national level, 15 regional data portals, where some of the national portals where community-supported
  • [2]
  • [3] For example to2006.rdf
< Prev   CONTENTS   Next >