BabelNet (http://babelnet.org) is a multilingual knowledge resource that
defines words and phrases in many languages. I've noticed that it copies
large amounts of content from Wikimedia projects, including Wikipedia,
Wiktionary, and Wikiquote, while violating Wikimedia's CC-By-SA license by
placing the content under an incompatible CC-By-NC-SA license.
As one example, I can search BabelNet for "Timsort", a Wikipedia article
whose first sentence is one I wrote:
The sentence I wrote appears at the top of the page (with credit to
Wikipedia). The rest of the page is also content remixed from Wikipedia,
including a gallery of images that are presented without credit. A scrolly
box in the footer of the page says the content is under the CC-By-NC-SA 3.0
license. Other pages, such as http://babelnet.org/synset?word=bn:00852566n,
combine data from multiple different resources.
The BabelNet creators are aware of the CC-By-SA licenses of the resources
they use (see http://babelnet.org/licenses/). In addition to the
non-commercial license they offer, their company, Babelscape (
http://babelscape.com/), sells commercial licenses to BabelNet.
I reached out to Roberto Navigli, who runs BabelNet and Babelscape, over
e-mail on March 23. I asked if the non-commercial license clause was simply
a mistake. In his reply, Navigli stated that BabelNet is not a derived
work, but is a CC-By-NC-SA-licensed collection made of several different
works. I responded that BabelNet doesn't meet the Creative Commons
definition of a "Collective Work", which would be necessary for it to not
be a derived work. Navigli responded:
"actually it is a collection of derivative work of several resources with
heretogeneous licenses, each of which clearly separated with separate
licenses and bundles. By transitivity derivative work is work with a
certain license, so it is work. Therefore, it is a collection of works with
different licenses and it can keep a separate license."
I believe this is nonsense on multiple levels. BabelNet is a derived work,
and if someone could disregard their obligation to share-alike their
derived work simply because they derived it from multiple resources, there
would be no point to putting ShareAlike clauses on data resources at all.
As a Wikipedia contributor (and a lapsed admin), I am sad to see BabelNet
appropriating the hard work of Wikimedians and others, placing a more
restrictive license on it, and selling it. This is also relevant for me
because I run ConceptNet (http://www.conceptnet.io/), a similar knowledge
resource, and I have made sure to follow Creative Commons license
requirements and to release all its data as CC-By-SA.
In a way I see BabelNet as a competitor, but ConceptNet is an open data
project and this space shouldn't have "competitors". If the Creative
Commons license were being used appropriately, then all of us working with
this kind of data would be collaborators in the world of Linked Open Data.
My preferred outcome would be to get BabelNet to change the copyright
notices and Creative Commons links on their site to remove the
"non-commercial" requirement, and to be able to download and use their data
under the CC-By-SA license that it should be under.
I'm sure Wikimedia has dealt with similar situations to this. What would be
the most effective next step to ensure that BabelNet follows the CC-By-SA
-- Rob Speer
Wikimedia-l mailing list, guidelines at:
New messages to: Wikimediaemail@example.com