Homogeneous licences - was Re: QUDT licensing & etc.

Hi.
I feel the need to comment that not all data from a source may benefit from the same licensing terms.

Some may remember the New York Times release of some data.
Included in there, were a lot of sameAs links to DBpedia.
But because those links were covered by the same terms as the rest of the (valuable) dataset, it was not possible for people to easily take them away and use them.
This was not in the NYT's interests: they really want people to take those triples away, so they can then bring traffic and eyes back to the dataset.

I think that in general, sameAs and similar links are probably best as something like CC0 for most data publishers (unless your business is generating sameAs-like knowledge, I suspect!).

Also, by the way, the provenance is likely to be different too, if there is provenance attached to the dataset; the NYT data was the product of intense curation over decades, but was in the same dataset as the work (sameAs reconciliation) of an intern which was a few months one summer.

And of course provenance metadata may well benefit from different terms to the dataset it relates to.

For example, I note (I think) that MusicBrainz has different dataset metadata for the raw music metadata and the other metadata that they infer from users' browsing.
So even when people take the music metadata away and use it (under their CC0 IIRC), it is still worth the consumer's while directing the users back to the MB site.

I'm not sure whether any of this applies in this case, but it is a question worth asking.

-- 
Hugh
023 8061 5652

Received on Wednesday, 16 September 2020 15:04:55 UTC