create wikidata importer for work

[martinantonmueller]

No description provided.

[csae8092]

@martinantonmueller could you maybe provide ~5 example wikidata uris per class

[martinantonmueller]

First I will try to find my way through a work, e.g. Romeo and Juliet, https://www.wikidata.org/wiki/Q83186

1. TITLE to get the title one does have to identify the language it is written in (P407) The result is a wikidata-item, in this case Q1860, "english". We need the abbreviation, so we have to look up In https://www.wikidata.org/wiki/Q1860 for the property P424, the language abbreviation. This is "en".
   With this result we go back to the entry https://www.wikidata.org/wiki/Q83186 and look for the label with the language code "en".
2. I would suggest the following fallback routine: If there is no language code then use "de" and if "de" is not available use "en"
3. DATE for the creation use inception and write the content of the first value of P571 without the time-stamp
4. no values for type of work etc.
5. AUTHOR create an author relation: use P50 to get the wikidata-id of the author. if that author is already in the PMB create a PMB-relation 1049 "hat geschaffen". if not first create the author by running the author-import script and then create the relation
6. store the wikidata-URI as URI

i wouldn't care too much for other details. one will have to go over the imported works anyways

[martinantonmueller]

i've separated the two importers. so here are five examples for works:

https://www.wikidata.org/wiki/Q1192140 (deutscher Text)
https://www.wikidata.org/wiki/Q1053387 (französischer Text)
https://www.wikidata.org/wiki/Q480 (spanischer Text)
https://www.wikidata.org/wiki/Q105745657 (deutscher Text)
https://www.wikidata.org/wiki/Q165318 (russischer Text, hier sollten kyrillische Zeichen importiert werden)

[csae8092]

given the fact that wikidata's datamodel for works is way to flexible/diverse; I'll only fetch wikidata-id from wikidata and the rest of the data (label, creators, creation date) from GND using -> pylobid.PylobidWork

@martinantonmueller GND sometimes provide two kinds of date: dateOfProduction and dateOfPublication wheras the first one often is some time span e.g. 1560-1566 and the later usually just a year.
In case both dates are provided by the GND which one should be saved?

[martinantonmueller]

dateofpublication! Thank you!

[martinantonmueller]

and if this would be a »Wunschkonzert« (wish concert) than the work would have dateofpublication and the relation to the author would have dateOfProduction. but the latter is not as important

[csae8092]

and if this would be a »Wunschkonzert« (wish concert) than the work would have dateofpublication and the relation to the author would have dateOfProduction. but the latter is not as important

x-mas is around the corner, let's see...
csae8092/pylobid#16

[martinantonmueller]

@csae8092 has anyone said you're incredible since I'm gone? i really hope so!

[csae8092]

ist nur schlechtes gewissen wegen der besch*** performance