DT3: Data licensing


Based on the discussion during the DCAT-AP workshop in Rome (13/05/2016), we extracted the following key points:

The current licence type vocabulary (http://purl.org/adms/licencetype/1.0) was established in agreement between W3C and the Commission’s legal services. However, this list was not based on practical considerations. It did not take into account the main questions that the users want to see answered, for example searching for open data versus non-open data, and did not try to match the types to the most commonly used licences, e.g. Creative Commons.

Agenzia per l’ Italia Digitale (AgID) tried to map the commonly used licences to the ADMS licence types but found out that it is difficult to establish a one-to-one correspondence. In some cases, it might be possible to link one licence to two ADMS licence types, but DCAT-AP only allows one licence type.

The European Data Portal (EDP)  has the problem that licence information is received from data providers in text and not in a normalised form, so they can get all kind of variants, e.g. “CC-by”, “cc-BY”, “Creative Commons Attribution”, or even with typos in the text. It is agreed by the group that the use of URIs, required by DCAT-AP, solves this issue. This allows the use of well-known URIs for common licence and may require the creation of URIs for specific local licences.

EDP uses its own type list and tries as much as possible to map incoming information, identifying gaps in the list, and then adding terms to it. Sometimes they need to contact the data providers to understand what a licence means. They are building an Excel table in which they map licences to their characteristics, e.g. identifying the permissions, obligations, restrictions and prohibitions a licence entails. Based on that analysis, they can also identify ‘similar’ licences. The Excel table can be shared.

A question was asked about the liability for the mapping, e.g. if you say that licence A is similar to licence B, and a user does something that is permitted under A but not under B, can they sue?

The user is responsible to verify the exact licence. The European Data Portal (EDP) does not replace the licence given for a particular piece of data. It is not a good practice to replace licence information. In some national environments (e.g. Austria) a single licence is mandated but it may be difficult to enforce that on a local level.

A mapping table could make clear that two URIs represent the same set of permissions and obligations, e.g.  https://creativecommons.org/cc0/ and http://publications.europa.eu/resource/authority/licence/CC0.

Identify licences: It was agreed that URIs should be used instead of text strings to identify licences, and that it would be good if there were tools for data providers that helped them to identify the right licence, e.g. with a drop-down list in the user interface that generates the right URI in the metadata.

Classify licences: It was agreed that the current ADMS licence type vocabulary is not fit for purpose. A simpler model should be considered for DCAT-AP, based on an analysis of user requirements. There could be a mix of a simple approach (open/not open) for the majority of cases and an additional, more detailed approach such as ODRL when necessary.

The participants agreed that DCAT-AP should provide more guidance on licensing.






Sun, 19/06/2016 - 12:51

Classifications etc. of licenses should not be considered normal metadata (maybe meta-metadata). Statements about licenses can and should be done elsewhere by linking licenses and classifications (using Linked Data).

Wed, 22/06/2016 - 15:31

Classification of licences is 'normal' metadata for the description of licences. Maybe you mean that the classification of licences is not metadata for the dataset. That is correct.

You are right that the classification of licences should be done elsewhere- However, when we were doing this work, we could not find anywhere where this is done in a citable way. The only licence classiification that we knew of was done for ADMS, so that is what DCAT-AP recommends.

Would you have any suggestion where we could find other, openly available classifications for licences?


Fri, 24/06/2016 - 19:59

Openly available classifications for licences can (should?) be provided by the Metadata Registry (MDR) of the Publications Office of the EU. The OKFN perhaps could do that as well. Another option would be the W3C Open Government Community Group (to which I belong).

Tue, 06/09/2016 - 16:02

The content of this field is kept private and will not be shown publicly.