Wikidata talk:Verifiability

Starting

I'll let others take the driving seat on the fluff, but hopefully the nutshell is a good summary of the community's position on the subject. —WFC— 11:06, 7 December 2012 (UTC)

It should at least be questionable if references (aka sources) do not support the statements. It could also be interesting to mark the references differently if they are sources (that is the value is the quote, primary source) or if they are references (a quote that qualifies the value, secondary source). Jeblad (talk) 07:56, 27 December 2012 (UTC)

Verifiability and low-structured sources

Verifiability is big problem for Wikidata. For example we have 5 statements sourced by some authoritative book. All is good, I can take the book and verify the statements manually. After some time we have 50 statements sourced by 10 different authoritative books. Situation becomes worse because I need high number of energy to verify the claims. 1000 statements and 200 books make the data unverifiable in practice. We have 65 000 000+ statements now. Current rule edition does not suggest any approaches to make this statement set verifiable. — Ivan A. Krestinin (talk) 04:27, 3 July 2015 (UTC)

Sources should be including page numbers. If they don't then they should. I'm not sure if that's a requirement we should have for this policy though.--Jasper Deng (talk) 05:07, 3 July 2015 (UTC)

Page numbers does not solve the problem. At least until artificial intelligence is not created. For example I need birth dates of all Russian scientists for my scientometric investigation. And I want to verify this dataset before usage. How many efforts is needed for this simple verification task in case of books and other classical sources? My raw estimate is one or two human-year. So this data subset is verifiable from formal point of view, but is not verifiable in practice. — Ivan A. Krestinin (talk) 20:19, 3 July 2015 (UTC)