21/07/2022 by Moissinac

My attention was recently drawn to the DBPedia license, which would imply that any dataset that relies on DBPedia should be made public with the same license. A stakeholder in a project told me that they had given up using DBPedia for this reason. According to the article https://fr.wikipedia.org/wiki/DBpedia, two licenses are applicable to DBPedia data; Creative Commons BY-SA 3.0 and GFDL.

I have found that other semantic datasets use the Creative Commons BY-SA 3.0 license. For example, the vocabularies proposed by the Ministry of Culture (http://data.culture.fr/thesaurus/static/en-savoir-plus) are released with this license.

This license stipulates in particular:

Share Under the Same Conditions - In the case that you remix, transform, or create from the material composing the original Work, you must distribute the modified Work under the same conditions, that is, with the same license with which the original Work was distributed.

It seems to me that creating many links to DBPedia uris in my own datasets implies that I am creating "from" DBPedia component material. The license seems to indicate that my data is a "mix" of DBPedia data, so it will have to be made public and accessible under the same license.

As an example, I had to deal with confidential data from an organization. This data had a geographic dimension regarding the origin of transactions. For this geographic data, we used URIs from DBpedia. The DBpedia license could imply that our data must be made public and reusable. Of course, in this simple case, a solution exists: create our own URIs for the geographic entities that are useful to us and isolate in a separate graph the sameAs links with the corresponding DBpedia URIs; only this last graph would be concerned by the license.

In conclusion, it seems necessary to be very careful when reusing sets of URIs proposed by the Linked Open Data (LOD).

Translated with www.DeepL.com/Translator (free version)