On 09/28/2015 08:30 AM, Daniel Kinzler wrote:
Am 28.09.2015 um 17:27 schrieb Peter F. Patel-Schneider:
Are you arguing against any tool that makes inferences combining multiple pieces of data in Wikidata? Would you also argue against this if the inferred information is flagged in some way?
I'm aguing against making such inference part of wikibase/wikidata core functionality, and hiding it's working ("magic").
However, I very much hope for a whole ecosystem of tools that apply and use such inference, and make the results obvious to users, both integrated with wikidata.org and outside.
Has anyone argued for performing inference and then hiding that it happened?
One problem that I see with Wikidata at the moment is that it is not obvious what inferences should or could be done. There is no theory of knowledge that stands behind Wikidata. It seems to me that in the absence of such a theory people are building bots that do some checks on the data in Wikipedia in an attempt to check some of the inferences that would be sanctioned by such a theory. However, I do not see any determination that the bots are covering those checks that should be made. I guess that now people are also building bots that make some of the inferences that would be sanctioned by a theory of knowledge for Wikidata. Again, however, there doesn't seem to any determination that these bots are making correct inferences or that they are covering a group of inferences that should be made.
I view this situation as inferior to an implementation of an integrated set of inferences for Wikidata.
Admittedly, coming up with a knowledge theory for Wikidata is not going to be easy. It is much easier to just write a bot that does something that might be useful.
peter