Hi everyone, next week I'll participate in a conference/hackathon called WikiCite [1]
It will be a great opportunity to talk about Wikidata and books in general, and of course I'd love to work about the integration of Wikidata and Wikisource.
It is my understanding (but please correct me if I'm wrong) that very few Wikisource integrated their books with Wikidata.
Of course, this is because the perennial issue of complicated relationships between books: a book as a "work", a book as an "edition" of that work. To complicate further, Wikisources host the ns0 textual version of a book and the nsIndex, but Wikidata do want only one sitelink...
Many discussions happened here [2], but I fear I'm lost now. I don't know where are we now, and if some community did indeed solve the problem. If your so, please tell :-)
It's really important.
Aubrey
[1] https://meta.wikimedia.org/wiki/WikiCite_2016 [2] https://www.wikidata.org/wiki/Wikidata_talk:Wikisource
Hello Andrea,
I don't believe that the integration state have move forward since Wikisource conference. The current state is something like: - if a work had many edition, have an item for the work and one for each edition, link the Wikisource disambiguation page to the work item and the main page of each edition (in ns0) to the edtion items. Do nothing for the index: page. - if a work had only one edition, have a single item for both and link to the main Page: of ns0 in Wikisource.
I would also be very happy to hear about if things moved in other communities.
Thomas
Le 21 mai 2016 à 10:48, Andrea Zanni zanni.andrea84@gmail.com a écrit :
Hi everyone, next week I'll participate in a conference/hackathon called WikiCite [1]
It will be a great opportunity to talk about Wikidata and books in general, and of course I'd love to work about the integration of Wikidata and Wikisource.
It is my understanding (but please correct me if I'm wrong) that very few Wikisource integrated their books with Wikidata.
Of course, this is because the perennial issue of complicated relationships between books: a book as a "work", a book as an "edition" of that work. To complicate further, Wikisources host the ns0 textual version of a book and the nsIndex, but Wikidata do want only one sitelink...
Many discussions happened here [2], but I fear I'm lost now. I don't know where are we now, and if some community did indeed solve the problem. If your so, please tell :-)
It's really important.
Aubrey
[1] https://meta.wikimedia.org/wiki/WikiCite_2016 [2] https://www.wikidata.org/wiki/Wikidata_talk:Wikisource _______________________________________________ Wikisource-l mailing list Wikisource-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikisource-l
Thanks Tpt for the quick response.
I feel that that is the situation for most Wikisources: the problem is that it's a static model, and not a procedure. For example: I put a new book in my Wikisource, what should I do? * check if is a new edition of a book in WS if not * check if there is already an item on WD ** if so, check if is work or edition ** if edition --> ok, link it ** if work, create edition, then link it
Workflows like this, I think, are not in place. We would need a WikidataWizard :-D
Aubrey
On Sat, May 21, 2016 at 11:13 AM, Thomas Tanon thomaspt@hotmail.fr wrote:
Hello Andrea,
I don't believe that the integration state have move forward since Wikisource conference. The current state is something like:
- if a work had many edition, have an item for the work and one for each
edition, link the Wikisource disambiguation page to the work item and the main page of each edition (in ns0) to the edtion items. Do nothing for the index: page.
- if a work had only one edition, have a single item for both and link to
the main Page: of ns0 in Wikisource.
I would also be very happy to hear about if things moved in other communities.
Thomas
Le 21 mai 2016 à 10:48, Andrea Zanni zanni.andrea84@gmail.com a écrit
:
Hi everyone, next week I'll participate in a conference/hackathon called WikiCite [1]
It will be a great opportunity to talk about Wikidata and books in
general, and of course I'd love to work about the integration of Wikidata and Wikisource.
It is my understanding (but please correct me if I'm wrong) that very
few Wikisource integrated their books with Wikidata.
Of course, this is because the perennial issue of complicated
relationships between books: a book as a "work", a book as an "edition" of that work.
To complicate further, Wikisources host the ns0 textual version of a
book and
the nsIndex, but Wikidata do want only one sitelink...
Many discussions happened here [2], but I fear I'm lost now. I don't know where are we now, and if some
community did indeed solve the problem.
If your so, please tell :-)
It's really important.
Aubrey
[1] https://meta.wikimedia.org/wiki/WikiCite_2016 [2] https://www.wikidata.org/wiki/Wikidata_talk:Wikisource _______________________________________________ Wikisource-l mailing list Wikisource-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikisource-l
Wikisource-l mailing list Wikisource-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikisource-l
2016-05-21 11:50 GMT+02:00 Andrea Zanni zanni.andrea84@gmail.com:
Thanks Tpt for the quick response.
I feel that that is the situation for most Wikisources: the problem is that it's a static model, and not a procedure.
Exactly what I tried to say : the problem is practical not theorical.
For example: I put a new book in my Wikisource, what should I do?
- check if is a new edition of a book in WS
if not
- check if there is already an item on WD
** if so, check if is work or edition ** if edition --> ok, link it ** if work, create edition, then link it
Workflows like this, I think, are not in place. We would need a WikidataWizard :-D
\o/ we definitely need that ! Who could build such a WonderfulWikidataWizard ?
Cdlt, ~nicolas
All this is way more complex than what we have at he.wikisource.
We only link the main ns0 page of a work, and usually we have only one edition of it. We don't use the index NS much, (since most of our books are either manually typed-in or OCRed privately and the text uploaded, either before or after having been proofread. If not proofread, we add a template).
This is the way our veteran users have gotten used to run things. I guess it drives would-be newcomers away. But we've been experimenting with a few works done with the Index:/Page: interface and it seems even more confusing to newcomers who don't know how to create a book from the proofread pages and see an accomplished result for their efforts.
On Sat, May 21, 2016 at 12:53 PM, Nicolas VIGNERON < vigneron.nicolas@gmail.com> wrote:
2016-05-21 11:50 GMT+02:00 Andrea Zanni zanni.andrea84@gmail.com:
Thanks Tpt for the quick response.
I feel that that is the situation for most Wikisources: the problem is that it's a static model, and not a procedure.
Exactly what I tried to say : the problem is practical not theorical.
For example: I put a new book in my Wikisource, what should I do?
- check if is a new edition of a book in WS
if not
- check if there is already an item on WD
** if so, check if is work or edition ** if edition --> ok, link it ** if work, create edition, then link it
Workflows like this, I think, are not in place. We would need a WikidataWizard :-D
\o/ we definitely need that ! Who could build such a WonderfulWikidataWizard ?
Cdlt, ~nicolas
Wikisource-l mailing list Wikisource-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikisource-l
The WikiCite conference is finished and we had lenghty discussions about Wikidata and books and citations. Some of that was useful also for Wikisource.
I'm trying to add some notes directly here: https://www.wikidata.org/wiki/Wikidata_talk:WikiProject_Books#Review_of_.22w...
But also I saw that en.source started doing some stuff with Wikidata, so I'm curious if there are some editors that can tell us something about it :-)
Aubrey
On Sun, May 22, 2016 at 1:14 AM, Nahum Wengrov novartza@gmail.com wrote:
All this is way more complex than what we have at he.wikisource.
We only link the main ns0 page of a work, and usually we have only one edition of it. We don't use the index NS much, (since most of our books are either manually typed-in or OCRed privately and the text uploaded, either before or after having been proofread. If not proofread, we add a template).
This is the way our veteran users have gotten used to run things. I guess it drives would-be newcomers away. But we've been experimenting with a few works done with the Index:/Page: interface and it seems even more confusing to newcomers who don't know how to create a book from the proofread pages and see an accomplished result for their efforts.
On Sat, May 21, 2016 at 12:53 PM, Nicolas VIGNERON < vigneron.nicolas@gmail.com> wrote:
2016-05-21 11:50 GMT+02:00 Andrea Zanni zanni.andrea84@gmail.com:
Thanks Tpt for the quick response.
I feel that that is the situation for most Wikisources: the problem is that it's a static model, and not a procedure.
Exactly what I tried to say : the problem is practical not theorical.
For example: I put a new book in my Wikisource, what should I do?
- check if is a new edition of a book in WS
if not
- check if there is already an item on WD
** if so, check if is work or edition ** if edition --> ok, link it ** if work, create edition, then link it
Workflows like this, I think, are not in place. We would need a WikidataWizard :-D
\o/ we definitely need that ! Who could build such a WonderfulWikidataWizard ?
Cdlt, ~nicolas
Wikisource-l mailing list Wikisource-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikisource-l
Wikisource-l mailing list Wikisource-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikisource-l
2016-05-21 10:48 GMT+02:00 Andrea Zanni zanni.andrea84@gmail.com:
Hi everyone, next week I'll participate in a conference/hackathon called WikiCite [1]
It will be a great opportunity to talk about Wikidata and books in general, and of course I'd love to work about the integration of Wikidata and Wikisource.
It is my understanding (but please correct me if I'm wrong) that very few Wikisource integrated their books with Wikidata.
Of course, this is because the perennial issue of complicated relationships between books: a book as a "work", a book as an "edition" of that work. To complicate further, Wikisources host the ns0 textual version of a book and the nsIndex, but Wikidata do want only one sitelink...
That's not really that complicated (see Tpt explanation and pages like https://www.wikidata.org/wiki/Wikidata:Wikisource/How_to_help ).
For the nsindex, there is a specific property : https://www.wikidata.org/wiki/Property:P1957
Many discussions happened here [2],
but I fear I'm lost now. I don't know where are we now, and if some community did indeed solve the problem. If your so, please tell :-)
It's really important.
I think the main problem is about pedagogy and easy-to-use tools (I love Wikidata but to be honest, the interface really sucks for newbies).
Cdlt, ~nicolas
BTW, in Rennes, we planned to think on June 18th about importing all the 13 millions books of the BnF catalog into Wikidata (or at least some part of it).
On Sat, May 21, 2016 at 11:41 AM, Nicolas VIGNERON < vigneron.nicolas@gmail.com> wrote:
BTW, in Rennes, we planned to think on June 18th about importing all the 13 millions books of the BnF catalog into Wikidata (or at least some part of it).
ok, we need to talk before that date :-D I personally think that it would be much easier to think about "federated" Wikibases, which can communicate with the real Wikidata in a second phase.
We were at least thinking this with the National library of Italy: it would allow them to control their data , but still have the possibility to have all the stuff in Wikidata at the same time. I personally think that integrations like this have 2 different issues: * one is technological: you have to map your existing metadata model to another * the second is social: you are entering a different community, with an existing culture and set of tools, practice.
For example, in WD we have property "title". Well, when I discussed with librarians in a bibliohackathon in June 2015, we argued a lot if that "title" was the same kind of "title" that they had in their catalog... Because there are several "titles" in MARC, for example.
Also, the moment that you map an existing catalog in Wikidata, you kind of lose information, and you can't come back. If you decide, for example, that all the different MARC titles go in property "title" in Wikidata, the moment you import it you don't know which was what, before the importing.
This is one of the reasons Cristian Bacchi designed a MARC-importer that is "lossless", and import MARC in a wikibase-based MARC set of properties...
Aubrey
wikisource-l@lists.wikimedia.org