I haven't seen BabelNet before. It seems to have similar goals to ConceptNet and a much more polished Web interface. It's pretty well done. And it seems to make use of Roberto Navigli's work on word sense disambiguation to align a bunch of different resources on the same set of senses.
But the biggest difference is that BabelNet adds the Non-Commercial restriction to its data. This has some interesting implications:
* You can use BabelNet for a research project, but if you succeed and want to use it in a real product -- even in an open-source tool that meets the OSI definition at
https://opensource.org/faq -- you have to negotiate a license with them.
* I think they're infringing Wikipedia's license by redistributing its content with added restrictions.
I don't think this is malicious on their part. I have found that academics often defensively add "non-commercial" restrictions to their work because they don't think about the implications, and because they're used to only working with other academics. It does concern me, though.