Here's the (nearly) equivalent query for the statements dump[1] loaded into Blazegraph:
PREFIX wd: http://www.wikidata.org/entity/ SELECT DISTINCT ?city ?citylabel ?mayorlabel WHERE { ?city wd:P31s/wd:P31v wd:Q515 . # find instances of subclasses of city ?city wd:P6s ?statement . # with a P6 (head of goverment) statement ?statement wd:P6v ?mayor . # ... that has the value ?mayor ?mayor wd:P21s/wd:P21v wd:Q6581072 . # ... where the ?mayor has P21 (sex or gender) female FILTER NOT EXISTS { ?statement wd:P582q ?x } # ... but the statement has no P582 (end date) qualifier
# Now select the population value of the ?city # (the number is reached through a chain of three properties) ?city wd:P1082s/wd:P1082v/http://www.wikidata.org/ontology#numericValue ?population .
# Optionally, find English labels for city and mayor: OPTIONAL { ?city wd:P373s/wd:P373v ?citylabel . # FILTER ( LANG(?citylabel) = "en" ) } OPTIONAL { ?mayor wd:P373s/wd:P373v ?mayorlabel . # FILTER ( LANG(?mayorlabel) = "en" ) }
} ORDER BY DESC(?population) LIMIT 100
Free beer to anyone who can figure out how to use those language filters. Would we need to also load property definitions[2]?
1. http://tools.wmflabs.org/wikidata-exports/rdf/exports/20150223/wikidata-stat... 2. http://tools.wmflabs.org/wikidata-exports/rdf/exports/20150223/wikidata-prop...
On Tue, Apr 21, 2015 at 11:13 AM, Jeremy Baron jeremy@tuxmachine.com wrote:
Hi,
On Tue, Apr 21, 2015 at 5:05 PM, Thad Guidry thadguidry@gmail.com wrote:
We had US Census, World Bank, and UN Data as our primary data sources
for any /statistics/ of a City/Town/Village. Here's Houston - https://www.freebase.com/m/03l2n#/location/statistical_region
I don't understand where a lot of those numbers are from.
Also, maybe Houston is a bad example because the Census Bureau revised numbers after the data was released.[0] Even some official Census Bureau sites still report the old, pre-appeal number.[1]
There are multiple years that have duplicate conflicting values after clicking "65 values total »" at your link. At first I was thinking it may be something like estimates base vs. estimate vs. decennial. However, for 2010 and 2011 there's one value that matches estimate from [1] (source = [2]) and a larger value (source = [3]) that does not match any other data I've seen. [2] and [3] both use the same "Attribution URI" [4].
In any case, why take this from freebase instead of importing directly from Census Bureau data? It's available in bulk. Format isn't great but isn't horrible either. (at least the 5-year ACS is inconsistent about upper/lower case for state two letter abbreviations. and, I think, most humans would prefer something like a geoid as a key rather than a dataset specific key used to look up the geoid in a different file. and other quirks)
-Jeremy
[0] http://www.chron.com/news/houston-texas/houston/article/City-wins-census-app... [1] http://factfinder.census.gov/bkmk/table/1.0/en/PEP/2013/PEPANNRES/1620000US4... [2] https://www.freebase.com/g/11x1k306j [3] https://www.freebase.com/m/0jst35z [4] http://www.census.gov/popest/about/terms.html
Wikidata-l mailing list Wikidata-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata-l