Thanks Ben for reading my mind, I was about to provide the same pointer. :-)
Let's try to keep the discussion on the primary sources tool in one
place as much as possible.
Cheers,
Marco
On 9/1/16 18:23, Benjamin Good wrote:
Dimitris,
This seems like good way to seed a large scale data and reference import
process. The trouble here is that wikidata already has large amounts of
such potentially useful data (e.g. most of freebase, the results of the
StepHit NLP system, etc.) but the processes for moving it in have thus
far gone slowly. In fact the author of the StepHit system for mining
facts/references for wikidata is shifting his focus entirely to
improving that part of the pipeline (known currently as the 'primary
sources' tool) as it is the bottleneck. It would be great to see you
get involved there:
https://www.wikidata.org/wiki/Wikidata:Requests_for_comment/Semi-automatic_…
Once we have a good technical and social pattern for verifying predicted
claims and references at scale, we can get to the business of loading
that system up with good input.
my two cents..
-ben
On Thu, Sep 1, 2016 at 7:53 AM, Dimitris Kontokostas <jimkont(a)gmail.com
<mailto:jimkont@gmail.com>> wrote:
Hmm,it is hard to interpret no feedback at all here, it could be
a) the data is not usable for Wikidata
b) this is not an interesting idea for Wikidata (now) or
c) this is not a good place to ask
Based on the very high activity on this list I could only guess (b),
even though but this idea came from the Wikidata community 1+ year
ago. This is probably not relevant now.
https://lists.wikimedia.org/pipermail/wikidata/2015-June/006366.html
<https://lists.wikimedia.org/pipermail/wikidata/2015-June/006366.html>
For reference, this is the prototype extractor that generated the
cited facts which can be run on newer dumps
https://github.com/dbpedia/extraction-framework/blob/master/core/src/main/s…
<https://github.com/dbpedia/extraction-framework/blob/master/core/src/main/scala/org/dbpedia/extraction/mappings/CitedFactsExtractor.scala>
Best,
Dimitris
On Tue, Aug 30, 2016 at 9:16 PM, Dario Taraborelli
<dtaraborelli(a)wikimedia.org <mailto:dtaraborelli@wikimedia.org>> wrote:
cc'ing wikicite-discuss, this is going to be of relevance to
many people there too.
On Mon, Aug 29, 2016 at 11:09 PM, Dimitris Kontokostas
<jimkont(a)gmail.com <mailto:jimkont@gmail.com>> wrote:
You can have a look here.
http://downloads.dbpedia.org/temporary/citations/enwiki-20160305-citedFacts…
<http://downloads.dbpedia.org/temporary/citations/enwiki-20160305-citedFacts.tql.bz2>
it is a quad file that contains DBpedia facts and I replaced
the context with the citation when the citation is on the
exact same line with the extracted fact. e.g.
<http://dbpedia.org/resource/An_American_in_Paris
<http://dbpedia.org/resource/An_American_in_Paris>>
<http://dbpedia.org/property/work
<http://dbpedia.org/property/work>> "An American in
Paris"@en
<https://www.bnote.de/?set=werk_detail&kompid=246&bnnr=16963&lc=en
<https://www.bnote.de/?set=werk_detail&kompid=246&bnnr=16963&lc=en>>
.
It is based on a complete English dump from ~April and
contains roughly 1M cited facts
This is more like a proof-of-concept and there are many ways
to improve and make it more usable for Wikidata
let me know what you think
On Mon, Aug 29, 2016 at 1:38 AM, Brill Lyle
<wp.brilllyle(a)gmail.com <mailto:wp.brilllyle@gmail.com>> wrote:
Yes? I think so. Except I would like to see fuller
citations extracted / sampled from / to? I don't have
the technical skill to understand the extraction
completely but Yes. I think there is very rich data in
Wikipedia that is very extractable.
Could this approach be a good candidate reference
suggestions in Wikidata?
(This particular one is already a reference but the
anthem and GDP in the attachment are not for example)
- Erika
*
*
*Erika Herzog*
Wikipedia *User:BrillLyle
<https://en.wikipedia.org/wiki/User:BrillLyle>*
On Sat, Aug 27, 2016 at 9:37 AM, Dimitris Kontokostas
<kontokostas(a)informatik.uni-leipzig.de
<mailto:kontokostas@informatik.uni-leipzig.de>> wrote:
Hi,
I had this idea for some time now but never got to
test/write it down.
DBpedia extracts detailed context information in
Quads (where possible) on where each triple came
from, including the line number in the wiki text.
Although each DBpedia extractor is independent,
using this context there is a small window for
combining output from different extractors, such as
the infobox statements we extract from Wikipedia and
the very recent citation extractors we announced [1]
I attach a very small sample from the article about
Germany where I filter out the related triples and
order them by the line number they were extracted
from e.g.
dbr:Germany dbo:populationTotal
"82175700"^^xsd:nonNegativeInteger
<http://en.wikipedia.org/wiki/Germany?oldid=736355524#
<http://en.wikipedia.org/wiki/Germany?oldid=736355524#>*absolute-line=66*&template=Infobox_country&property=population_estimate&split=1&wikiTextSize=10&plainTextSize=10&valueSize=8>
.
<https://www.destatis.de/DE/PresseService/Presse/Pressemitteilungen/2016/08/PD16_295_12411pdf.pdf;jsessionid=996EC2DF0A8D510CF89FDCBC74DBAE9F.cae2?__blob=publicationFile
<https://www.destatis.de/DE/PresseService/Presse/Pressemitteilungen/2016/08/PD16_295_12411pdf.pdf;jsessionid=996EC2DF0A8D510CF89FDCBC74DBAE9F.cae2?__blob=publicationFile>>
dbp:isCitedBy dbr:Germany
<http://en.wikipedia.org/wiki/Germany?oldid=736355524#
<http://en.wikipedia.org/wiki/Germany?oldid=736355524#>*absolute-line=66*>
.
Looking at the wikipedia article we see:
|population_estimate = 82,175,700<ref>{{cite
web|url=https://www.destatis.de/DE/PresseService/Presse/Pressemitteilungen/2016/08/PD16_295_12411pdf.pdf;jsessionid=996EC2DF0A8D510CF89FDCBC74DBAE9F.cae2?__blob=publicationFile|title=Population
<https://www.destatis.de/DE/PresseService/Presse/Pressemitteilungen/2016/08/PD16_295_12411pdf.pdf;jsessionid=996EC2DF0A8D510CF89FDCBC74DBAE9F.cae2?__blob=publicationFile%7Ctitle=Population>
at 82.2 million at the end of 2015 – population
increase due to high immigration|date=26 August
2016|work=destatis.de <http://destatis.de>}}</ref>
Could this approach be a good candidate reference
suggestions in Wikidata?
(This particular one is already a reference but the
anthem and GDP in the attachment are not for example)
There are many things that can be done to improve
the matching but before getting into details I would
like to see if this idea is worth exploring more or not
Cheers,
Dimitris
[1]
http://www.mail-archive.com/dbpedia-discussion%40lists.sourceforge.net/msg0…
<http://www.mail-archive.com/dbpedia-discussion%40lists.sourceforge.net/msg07739.html>
--
Dimitris Kontokostas
Department of Computer Science, University of
Leipzig & DBpedia Association
Projects:
http://dbpedia.org,
http://rdfunit.aksw.org,
http://aligned-project.eu
Homepage:
http://aksw.org/DimitrisKontokostas
<http://aksw.org/DimitrisKontokostas>
Research Group:
AKSW/KILT
http://aksw.org/Groups/KILT
<http://aksw.org/Groups/KILT>
_______________________________________________
Wikidata mailing list
Wikidata(a)lists.wikimedia.org
<mailto:Wikidata@lists.wikimedia.org>
https://lists.wikimedia.org/mailman/listinfo/wikidata
<https://lists.wikimedia.org/mailman/listinfo/wikidata>
_______________________________________________
Wikidata mailing list
Wikidata(a)lists.wikimedia.org
<mailto:Wikidata@lists.wikimedia.org>
https://lists.wikimedia.org/mailman/listinfo/wikidata
<https://lists.wikimedia.org/mailman/listinfo/wikidata>
--
Kontokostas Dimitris
_______________________________________________
Wikidata mailing list
Wikidata(a)lists.wikimedia.org
<mailto:Wikidata@lists.wikimedia.org>
https://lists.wikimedia.org/mailman/listinfo/wikidata
<https://lists.wikimedia.org/mailman/listinfo/wikidata>
--
*Dario Taraborelli *Head of Research, Wikimedia Foundation
wikimediafoundation.org
<http://wikimediafoundation.org/> •
nitens.org
<http://nitens.org/> • @readermeter
<http://twitter.com/readermeter>
--
WikiCite 2016 – May 26-26, 2016, Berlin
Meta:
https://meta.wikimedia.org/wiki/WikiCite_2016
<https://meta.wikimedia.org/wiki/WikiCite_2016>
Twitter:
https://twitter.com/wikicite16
---
You received this message because you are subscribed to the
Google Groups "wikicite-discuss" group.
To unsubscribe from this group and stop receiving emails from
it, send an email to wikicite-discuss+unsubscribe(a)wikimedia.org
<mailto:wikicite-discuss+unsubscribe@wikimedia.org>.
--
Kontokostas Dimitris
_______________________________________________
Wikidata mailing list
Wikidata(a)lists.wikimedia.org <mailto:Wikidata@lists.wikimedia.org>
https://lists.wikimedia.org/mailman/listinfo/wikidata
<https://lists.wikimedia.org/mailman/listinfo/wikidata>
_______________________________________________
Wikidata mailing list
Wikidata(a)lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikidata