Wikidata query request - Cloud - lists.wikimedia.org

Jaime Crespo

13 Mar 13 Mar

9:39 a.m.

I am not 100% sure there is a perfect way to do what you want by querying the metadata databases (I assume that is what you mean with query)- I don't think that data is metadata, but content itself, which is not on the metadata databases. Calling the wikidata query service is probably what you want: < https://query.wikidata.org/#SELECT%20%3Fitem%20%3FitemLabel%20%3Fgeoname%0A…

...

Note the LIMIT and OFFSET that will let you iterate over the dataset (a where close would be faster). There is a way to get results, which is iterating over: < https://www.wikidata.org/w/index.php?title=Special:WhatLinksHere/Property:P…

...

That is a standard mediawiki api query, you will also find this on the pagelinks table, but you should check every page you get afterwards (by retrieving its contents), as it could include false positives or be behind on updates. On Sun, Mar 11, 2018 at 3:44 PM, Huji Lee <huji.huji(a)gmail.com> wrote: > Hello,

...

> I need help writing a query that I would like to run on the Clouds. The > goal of the query is to retrieve the following information from > wikidatawiki_p:

...

> * Find all pages that have a claim for the property P1566, for example see > https://www.wikidata.org/wiki/Q2113430 > * Find out what is the value of their P1566 property (in this case, 18918)

...

> Output format should be like this:

...

> page_id entity property_value > 2039804 Q2113430 18918 > ...

...

> Thanks in advance,

...

> Huji

...

> _______________________________________________ > Wikimedia Cloud Services mailing list > Cloud(a)lists.wikimedia.org (formerly labs-l(a)lists.wikimedia.org) > https://lists.wikimedia.org/mailman/listinfo/cloud

...

-- Jaime Crespo <http://wikimedia.org

...

Reply

Lucas Werkmeister

11 a.m.

Huji, do you need the page_id in the query results? Otherwise, I would suggest using either the Wikidata Query Service, as Jaime suggested (though I’d omit the LIMIT and OFFSET – I think it’s better to let the server send you all the results at once) or the Linked Data Fragments server: https://query.wikidata.org/bigdata/ldf?subject=&predicate=http%3A%2F%2F… (this URL will return HTML, RDF XML, Turtle, LD-JSON, … depending on Accept header). Cheers, Lucas 2018-03-14 1:03 GMT+01:00 Huji Lee <huji.huji(a)gmail.com>om>:

...

Thanks, Jaime, for your recommendation. If I understand the result of [1] correctly, there are around 3.5 million pages with a GeoNames property specified on Wikidata. I'm sure some of them are redirects, or not cities, etc. But still, going through millions of pages through API calls of 1000 at a time is cumbersome and inefficient. (The example you gave takes 20 seconds to run; that would mean a total of 20 * 3.5 * 1000 seconds which is like 19 hours, assuming no lag or error). However, what you suggested gave me an idea: I can take a look at the code for the Api itself (I guess it is at [2]) and figure out how the query is written there, then try to write a similar query on my own. If I figure it out, I will report back here. Huji [1] https://quarry.wmflabs.org/query/25418 [2] https://phabricator.wikimedia.org/diffusion/EWBA/browse/ master/client/includes/Api/ApiPropsEntityUsage.php On Tue, Mar 13, 2018 at 5:39 AM, Jaime Crespo <jcrespo(a)wikimedia.org> wrote:

I am not 100% sure there is a perfect way to do what you want by querying the metadata databases (I assume that is what you mean with query)- I don't think that data is metadata, but content itself, which is not on the metadata databases. Calling the wikidata query service is probably what you want: <https://query.wikidata.org/#SELECT%20%3Fitem%20%3FitemLabel %20%3Fgeoname%0AWHERE%20%7B%0A%09%3Fitem%20wdt%3AP1566%20% 3Fgeoname%20.%0A%09SERVICE%20wikibase%3Alabel%20%7B%20bd% 3AserviceParam%20wikibase%3Alanguage%20%22%5BAUTO_LANGUAGE% 5D%2Cen%22%20%7D%0A%7D%0ALIMIT%201000%20OFFSET%201000> Note the LIMIT and OFFSET that will let you iterate over the dataset (a where close would be faster). There is a way to get results, which is iterating over: <https://www.wikidata.org/w/index.php?title=Special:WhatLink sHere/Property:P1566&hidetrans=1&hideredirs=1> That is a standard mediawiki api query, you will also find this on the pagelinks table, but you should check every page you get afterwards (by retrieving its contents), as it could include false positives or be behind on updates. On Sun, Mar 11, 2018 at 3:44 PM, Huji Lee <huji.huji(a)gmail.com> wrote:

Hello, I need help writing a query that I would like to run on the Clouds. The goal of the query is to retrieve the following information from wikidatawiki_p: * Find all pages that have a claim for the property P1566, for example see https://www.wikidata.org/wiki/Q2113430 * Find out what is the value of their P1566 property (in this case, 18918) Output format should be like this: page_id entity property_value 2039804 Q2113430 18918 ... Thanks in advance, Huji _______________________________________________ Wikimedia Cloud Services mailing list Cloud(a)lists.wikimedia.org (formerly labs-l(a)lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/cloud

-- Jaime Crespo <http://wikimedia.org> _______________________________________________ Wikimedia Cloud Services mailing list Cloud(a)lists.wikimedia.org (formerly labs-l(a)lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/cloud

_______________________________________________ Wikimedia Cloud Services mailing list Cloud(a)lists.wikimedia.org (formerly labs-l(a)lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/cloud

-- Lucas Werkmeister Software Developer (Intern) Wikimedia Deutschland e. V. | Tempelhofer Ufer 23-24 | 10963 Berlin Phone: +49 (0)30 219 158 26-0 https://wikimedia.de Imagine a world, in which every single human being can freely share in the sum of all knowledge. That‘s our commitment. Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e. V. Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg unter der Nummer 23855 B. Als gemeinnützig anerkannt durch das Finanzamt für Körperschaften I Berlin, Steuernummer 27/029/42207.

Reply

Huji Lee

4:34 p.m.

Actually, never mind. I reviewed the Java code behind it and it doesn't support more items per page. It also gets slow when you look at later pages (first few pages are in a warm cache and are fast). I think my best bet is to just download the latest JSON dump from https://www.wikidata.org/wiki/Wikidata:Database_download and parse it myself. Thanks again! Huji On Wed, Mar 14, 2018 at 12:12 PM, Huji Lee <huji.huji(a)gmail.com> wrote:

...

Lucas, No I don't need the page_id. The other two are enough. Wikidata Query Service seems very slow (it'll take about one day of continuous querying to get all the data). Linked Data Fragments server seems faster, but I wish I knew how to make it return more than 100 results at a time. Do you? Thanks, Huji On Wed, Mar 14, 2018 at 7:00 AM, Lucas Werkmeister < lucas.werkmeister(a)wikimedia.de> wrote:

Huji, do you need the page_id in the query results? Otherwise, I would suggest using either the Wikidata Query Service, as Jaime suggested (though I’d omit the LIMIT and OFFSET – I think it’s better to let the server send you all the results at once) or the Linked Data Fragments server: https://query.wikidata.org/bigdata/ldf?subject=&predicate= http%3A%2F%2Fwww.wikidata.org%2Fprop%2Fdirect%2FP1566&object= (this URL will return HTML, RDF XML, Turtle, LD-JSON, … depending on Accept header). Cheers, Lucas 2018-03-14 1:03 GMT+01:00 Huji Lee <huji.huji(a)gmail.com>om>:

Thanks, Jaime, for your recommendation. If I understand the result of [1] correctly, there are around 3.5 million pages with a GeoNames property specified on Wikidata. I'm sure some of them are redirects, or not cities, etc. But still, going through millions of pages through API calls of 1000 at a time is cumbersome and inefficient. (The example you gave takes 20 seconds to run; that would mean a total of 20 * 3.5 * 1000 seconds which is like 19 hours, assuming no lag or error). However, what you suggested gave me an idea: I can take a look at the code for the Api itself (I guess it is at [2]) and figure out how the query is written there, then try to write a similar query on my own. If I figure it out, I will report back here. Huji [1] https://quarry.wmflabs.org/query/25418 [2] https://phabricator.wikimedia.org/diffusion/EWBA/browse/mast er/client/includes/Api/ApiPropsEntityUsage.php On Tue, Mar 13, 2018 at 5:39 AM, Jaime Crespo <jcrespo(a)wikimedia.org> wrote:

I am not 100% sure there is a perfect way to do what you want by querying the metadata databases (I assume that is what you mean with query)- I don't think that data is metadata, but content itself, which is not on the metadata databases. Calling the wikidata query service is probably what you want: <https://query.wikidata.org/#SELECT%20%3Fitem%20%3FitemLabel %20%3Fgeoname%0AWHERE%20%7B%0A%09%3Fitem%20wdt%3AP1566%20%3F geoname%20.%0A%09SERVICE%20wikibase%3Alabel%20%7B%20bd%3Aser viceParam%20wikibase%3Alanguage%20%22%5BAUTO_LANGUAGE%5D%2Ce n%22%20%7D%0A%7D%0ALIMIT%201000%20OFFSET%201000> Note the LIMIT and OFFSET that will let you iterate over the dataset (a where close would be faster). There is a way to get results, which is iterating over: <https://www.wikidata.org/w/index.php?title=Special:WhatLink sHere/Property:P1566&hidetrans=1&hideredirs=1> That is a standard mediawiki api query, you will also find this on the pagelinks table, but you should check every page you get afterwards (by retrieving its contents), as it could include false positives or be behind on updates. On Sun, Mar 11, 2018 at 3:44 PM, Huji Lee <huji.huji(a)gmail.com> wrote: > Hello, > > I need help writing a query that I would like to run on the Clouds. > The goal of the query is to retrieve the following information from > wikidatawiki_p: > > * Find all pages that have a claim for the property P1566, for example > see https://www.wikidata.org/wiki/Q2113430 > * Find out what is the value of their P1566 property (in this case, > 18918) > > Output format should be like this: > > page_id entity property_value > 2039804 Q2113430 18918 > ... > > Thanks in advance, > > Huji > > _______________________________________________ > Wikimedia Cloud Services mailing list > Cloud(a)lists.wikimedia.org (formerly labs-l(a)lists.wikimedia.org) > https://lists.wikimedia.org/mailman/listinfo/cloud > -- Jaime Crespo <http://wikimedia.org> _______________________________________________ Wikimedia Cloud Services mailing list Cloud(a)lists.wikimedia.org (formerly labs-l(a)lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/cloud

_______________________________________________ Wikimedia Cloud Services mailing list Cloud(a)lists.wikimedia.org (formerly labs-l(a)lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/cloud

-- Lucas Werkmeister Software Developer (Intern) Wikimedia Deutschland e. V. | Tempelhofer Ufer 23-24 | 10963 Berlin Phone: +49 (0)30 219 158 26-0 https://wikimedia.de Imagine a world, in which every single human being can freely share in the sum of all knowledge. That‘s our commitment. Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e. V. Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg unter der Nummer 23855 B. Als gemeinnützig anerkannt durch das Finanzamt für Körperschaften I Berlin, Steuernummer 27/029/42207. _______________________________________________ Wikimedia Cloud Services mailing list Cloud(a)lists.wikimedia.org (formerly labs-l(a)lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/cloud

Reply

Huji Lee

8:01 p.m.

Seems like it was a network latency issue on my part. Changing the network made it possible for me to download it like you did, in around 3.5 minutes. I'm almost done processing the data! Thanks so much, Huji On Wed, Mar 14, 2018 at 12:39 PM, Lucas Werkmeister < lucas.werkmeister(a)wikimedia.de> wrote:

...

Sorry, I’m not sure what you mean about the query service being slow? I was able to fetch all 3.5M results with the following command: $ time curl --silent --header 'Accept: application/json' --get --data-urlencode 'query=SELECT ?entity ?property_value WHERE { ?entity wdt:P1566 ?property_value. }' https://query.wikidata.org/sparql | xz > /tmp/geonames.xz real 2m28,161s user 2m27,023s sys 0m4,334s $ time unxz < /tmp/geonames.xz | jq '.results.bindings | length' 3495429 real 0m33,030s user 0m33,467s sys 0m4,702s Sure, it took a few minutes to download, but that’s not unexpected for a large result set :) but as long as you don’t use LIMIT and OFFSET, the query is quite efficient, and it’s only bound by how fast the query service can pump out the data. (If it was really computationally expensive, it would be killed after sixty seconds anyways.) 2018-03-14 17:34 GMT+01:00 Huji Lee <huji.huji(a)gmail.com>om>:

Actually, never mind. I reviewed the Java code behind it and it doesn't support more items per page. It also gets slow when you look at later pages (first few pages are in a warm cache and are fast). I think my best bet is to just download the latest JSON dump from https://www.wikidata.org/wiki/Wikidata:Database_download and parse it myself. Thanks again! Huji On Wed, Mar 14, 2018 at 12:12 PM, Huji Lee <huji.huji(a)gmail.com> wrote:

Lucas, No I don't need the page_id. The other two are enough. Wikidata Query Service seems very slow (it'll take about one day of continuous querying to get all the data). Linked Data Fragments server seems faster, but I wish I knew how to make it return more than 100 results at a time. Do you? Thanks, Huji On Wed, Mar 14, 2018 at 7:00 AM, Lucas Werkmeister < lucas.werkmeister(a)wikimedia.de> wrote:

Huji, do you need the page_id in the query results? Otherwise, I would suggest using either the Wikidata Query Service, as Jaime suggested (though I’d omit the LIMIT and OFFSET – I think it’s better to let the server send you all the results at once) or the Linked Data Fragments server: https://query.wikidata.org/bigdata/ldf?subject=&predicate=ht tp%3A%2F%2Fwww.wikidata.org%2Fprop%2Fdirect%2FP1566&object= (this URL will return HTML, RDF XML, Turtle, LD-JSON, … depending on Accept header). Cheers, Lucas 2018-03-14 1:03 GMT+01:00 Huji Lee <huji.huji(a)gmail.com>om>: > Thanks, Jaime, for your recommendation. > > If I understand the result of [1] correctly, there are around 3.5 > million pages with a GeoNames property specified on Wikidata. I'm sure some > of them are redirects, or not cities, etc. But still, going through > millions of pages through API calls of 1000 at a time is cumbersome and > inefficient. (The example you gave takes 20 seconds to run; that would mean > a total of 20 * 3.5 * 1000 seconds which is like 19 hours, assuming no lag > or error). > > However, what you suggested gave me an idea: I can take a look at the > code for the Api itself (I guess it is at [2]) and figure out how the query > is written there, then try to write a similar query on my own. If I figure > it out, I will report back here. > > Huji > > [1] https://quarry.wmflabs.org/query/25418 > [2] https://phabricator.wikimedia.org/diffusion/EWBA/browse/mast > er/client/includes/Api/ApiPropsEntityUsage.php > > On Tue, Mar 13, 2018 at 5:39 AM, Jaime Crespo <jcrespo(a)wikimedia.org> > wrote: > >> I am not 100% sure there is a perfect way to do what you want by >> querying the metadata databases (I assume that is what you mean with >> query)- I don't think that data is metadata, but content itself, which is >> not on the metadata databases. >> >> Calling the wikidata query service is probably what you want: >> >> <https://query.wikidata.org/#SELECT%20%3Fitem%20%3FitemLabel >> %20%3Fgeoname%0AWHERE%20%7B%0A%09%3Fitem%20wdt%3AP1566%20%3F >> geoname%20.%0A%09SERVICE%20wikibase%3Alabel%20%7B%20bd%3Aser >> viceParam%20wikibase%3Alanguage%20%22%5BAUTO_LANGUAGE%5D%2Ce >> n%22%20%7D%0A%7D%0ALIMIT%201000%20OFFSET%201000> >> >> Note the LIMIT and OFFSET that will let you iterate over the dataset >> (a where close would be faster). >> >> There is a way to get results, which is iterating over: >> <https://www.wikidata.org/w/index.php?title=Special:WhatLink >> sHere/Property:P1566&hidetrans=1&hideredirs=1> >> >> That is a standard mediawiki api query, you will also find this on >> the pagelinks table, but you should check every page you get afterwards (by >> retrieving its contents), as it could include false positives or be behind >> on updates. >> >> On Sun, Mar 11, 2018 at 3:44 PM, Huji Lee <huji.huji(a)gmail.com> >> wrote: >> >>> Hello, >>> >>> I need help writing a query that I would like to run on the Clouds. >>> The goal of the query is to retrieve the following information from >>> wikidatawiki_p: >>> >>> * Find all pages that have a claim for the property P1566, for >>> example see https://www.wikidata.org/wiki/Q2113430 >>> * Find out what is the value of their P1566 property (in this case, >>> 18918) >>> >>> Output format should be like this: >>> >>> page_id entity property_value >>> 2039804 Q2113430 18918 >>> ... >>> >>> Thanks in advance, >>> >>> Huji >>> >>> _______________________________________________ >>> Wikimedia Cloud Services mailing list >>> Cloud(a)lists.wikimedia.org (formerly labs-l(a)lists.wikimedia.org) >>> https://lists.wikimedia.org/mailman/listinfo/cloud >>> >> >> >> >> -- >> Jaime Crespo >> <http://wikimedia.org> >> >> _______________________________________________ >> Wikimedia Cloud Services mailing list >> Cloud(a)lists.wikimedia.org (formerly labs-l(a)lists.wikimedia.org) >> https://lists.wikimedia.org/mailman/listinfo/cloud >> > > > _______________________________________________ > Wikimedia Cloud Services mailing list > Cloud(a)lists.wikimedia.org (formerly labs-l(a)lists.wikimedia.org) > https://lists.wikimedia.org/mailman/listinfo/cloud > -- Lucas Werkmeister Software Developer (Intern) Wikimedia Deutschland e. V. | Tempelhofer Ufer 23-24 | 10963 Berlin Phone: +49 (0)30 219 158 26-0 https://wikimedia.de Imagine a world, in which every single human being can freely share in the sum of all knowledge. That‘s our commitment. Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e. V. Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg unter der Nummer 23855 B. Als gemeinnützig anerkannt durch das Finanzamt für Körperschaften I Berlin, Steuernummer 27/029/42207. _______________________________________________ Wikimedia Cloud Services mailing list Cloud(a)lists.wikimedia.org (formerly labs-l(a)lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/cloud

_______________________________________________ Wikimedia Cloud Services mailing list Cloud(a)lists.wikimedia.org (formerly labs-l(a)lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/cloud

-- Lucas Werkmeister Software Developer (Intern) Wikimedia Deutschland e. V. | Tempelhofer Ufer 23-24 | 10963 Berlin Phone: +49 (0)30 219 158 26-0 https://wikimedia.de Imagine a world, in which every single human being can freely share in the sum of all knowledge. That‘s our commitment. Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e. V. Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg unter der Nummer 23855 B. Als gemeinnützig anerkannt durch das Finanzamt für Körperschaften I Berlin, Steuernummer 27/029/42207. _______________________________________________ Wikimedia Cloud Services mailing list Cloud(a)lists.wikimedia.org (formerly labs-l(a)lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/cloud

Reply