wikibase:directClaim predicate? - Wikidata - lists.wikimedia.org

List overview All Threads
Download

wikibase:directClaim predicate?

Fwd: [Wiki-research-l] Research...

Weekly Summary #253

Bob DuCharme

18 Mar 2017 18 Mar '17

6:03 p.m.

Could someone tell me what this means? http://wikiba.se/ontology-1.0.owl#directClaim describes it as a "Link between Wikibase Property and direct claim predicate" and the definition in https://www.mediawiki.org/wiki/Wikibase/Indexing/RDF_Dump_Format is similar. What makes a predicate a direct claim predicate? Thanks, Bob

Reply

Show replies by date

Daniel Kinzler

18 Mar 18 Mar

7:52 p.m.

Am 18.03.2017 um 19:03 schrieb Bob DuCharme:

What makes a predicate a direct claim predicate?

It's a predicate (that's what RDF calls all relationships) that expresses a direct claim (as opposed to a full statement). Direct claims are one of two ways Wikidata Statements are mapped to RDF. In the wikidata query service, each statement is represented twice - once as a full statement, and once as a direct claim. Direct claims represent a "naive projection" of wikidata to RDF: everything that is claimed (by anyone) to be true (under any circumstances) is assumed to be true. So you get triples like <wd:Q64> <wdt:P1376> <wd:Q183> meaning "Berlin - capital-of - Germany". Simple to work with, but incomplete: you also get <wd:Q64> <wdt:P1376> <wd:Q27306> ("Berlin - capital-of - Kingdom of Prussia"), without an easy way to see that one is current and the other is not. To get all the additional context information, you need to look at the full statement representation, which provides a complex structure of value, qualifiers, and source references. The full mapping will use predicates with the "wds" prefix to connect the item (subject) to the structure representing the statement with all its parts. -- Daniel Kinzler Senior Software Developer Wikimedia Deutschland Gesellschaft zur Förderung Freien Wissens e.V.

Reply

Bob DuCharme

8:27 p.m.

Thanks Daniel! How do I find a full statement representation? For example, what would the full statement representation be for a triple like {wd:Q64 wdt:P1376 wd:Q183}? Bob On 03/18/2017 03:52 PM, Daniel Kinzler wrote:

Am 18.03.2017 um 19:03 schrieb Bob DuCharme:

What makes a predicate a direct claim predicate?

It's a predicate (that's what RDF calls all relationships) that expresses a direct claim (as opposed to a full statement). Direct claims are one of two ways Wikidata Statements are mapped to RDF. In the wikidata query service, each statement is represented twice - once as a full statement, and once as a direct claim. Direct claims represent a "naive projection" of wikidata to RDF: everything that is claimed (by anyone) to be true (under any circumstances) is assumed to be true. So you get triples like <wd:Q64> <wdt:P1376> <wd:Q183> meaning "Berlin - capital-of - Germany". Simple to work with, but incomplete: you also get <wd:Q64> <wdt:P1376> <wd:Q27306> ("Berlin - capital-of - Kingdom of Prussia"), without an easy way to see that one is current and the other is not. To get all the additional context information, you need to look at the full statement representation, which provides a complex structure of value, qualifiers, and source references. The full mapping will use predicates with the "wds" prefix to connect the item (subject) to the structure representing the statement with all its parts.

Reply

Daniel Kinzler

9:27 p.m.

Am 18.03.2017 um 21:27 schrieb Bob DuCharme:

Thanks Daniel! How do I find a full statement representation? For example, what would the full statement representation be for a triple like {wd:Q64 wdt:P1376 wd:Q183}?

The full representation of the statement in this case is: wds:Q64-43CCD3D6-F52E-4742-B0E3-BCA671B69D2C a wikibase:Statement, wikibase:BestRank ; wikibase:rank wikibase:PreferredRank ; ps:P1376 wd:Q183 ; prov:wasDerivedFrom wdref:ba76a7c0f885fa85b10368696ab4ac89680aa073 . wdref:ba76a7c0f885fa85b10368696ab4ac89680aa073 a wikibase:Reference ; pr:P248 wd:Q451546 ; pr:P958 "Artikel 2 (1)" . This RDF representation can be found at <https://www.wikidata.org/wiki/Special:EntityData/Q64.ttl>. Content negotiation will take you there from the canonical URI, <https://www.wikidata.org/entity/Q64.ttl> In addition to the actual value, the RDF above also give the rank, and a source reference (nameley, the re-unification treaty). This statement doesn't currently have a qualifier - it should have at least one, stating since when Berlin is the Capital of Germany. That qualifier would be represented as: wds:Q64-43CCD3D6-F52E-4742-B0E3-BCA671B69D2C pq:P580 "1990-10-03T00:00:00Z"^^xsd:dateTime ; The Statement ID, Q64$43CCD3D6-F52E-4742-B0E3-BCA671B69D2C, can be found in the HTML source of the page, encoded as a CSS class. These IDs are not exposed nicely anywhere. But usually, one would look at the RDF representation right away, or at least got from HTML to *all* the RDF. HTH -- daniel -- Daniel Kinzler Senior Software Developer Wikimedia Deutschland Gesellschaft zur Förderung Freien Wissens e.V.

Reply

Bob DuCharme

9:48 p.m.

Before I had a difficult time understanding https://www.mediawiki.org/wiki/Wikibase/Indexing/RDF_Dump_Format, but with that full example of https://www.wikidata.org/wiki/Special:EntityData/Q84.ttl in front of me, a lot of the RDF Dump Format page is now easier for me to understand. New question: when I see that https://www.wikidata.org/wiki/Special:EntityData says "This page provides a linked data interface to entity values", can you tell me what "entity" means in the context of Wikidata? If I was going to refer to something that can be identified with a URI and described by triples in which it is the subject, I would just use the term "resource" as described at https://www.w3.org/TR/rdf11-concepts/#resources-and-statements (and remembering what "RDF" stands for!) so I'm guessing that "entity" means something a little more specific than that here. Thanks, Bob On 03/18/2017 05:27 PM, Daniel Kinzler wrote:

Am 18.03.2017 um 21:27 schrieb Bob DuCharme:

Thanks Daniel! How do I find a full statement representation? For example, what would the full statement representation be for a triple like {wd:Q64 wdt:P1376 wd:Q183}?

The full representation of the statement in this case is: wds:Q64-43CCD3D6-F52E-4742-B0E3-BCA671B69D2C a wikibase:Statement, wikibase:BestRank ; wikibase:rank wikibase:PreferredRank ; ps:P1376 wd:Q183 ; prov:wasDerivedFrom wdref:ba76a7c0f885fa85b10368696ab4ac89680aa073 . wdref:ba76a7c0f885fa85b10368696ab4ac89680aa073 a wikibase:Reference ; pr:P248 wd:Q451546 ; pr:P958 "Artikel 2 (1)" . This RDF representation can be found at <https://www.wikidata.org/wiki/Special:EntityData/Q64.ttl>. Content negotiation will take you there from the canonical URI, <https://www.wikidata.org/entity/Q64.ttl> In addition to the actual value, the RDF above also give the rank, and a source reference (nameley, the re-unification treaty). This statement doesn't currently have a qualifier - it should have at least one, stating since when Berlin is the Capital of Germany. That qualifier would be represented as: wds:Q64-43CCD3D6-F52E-4742-B0E3-BCA671B69D2C pq:P580 "1990-10-03T00:00:00Z"^^xsd:dateTime ; The Statement ID, Q64$43CCD3D6-F52E-4742-B0E3-BCA671B69D2C, can be found in the HTML source of the page, encoded as a CSS class. These IDs are not exposed nicely anywhere. But usually, one would look at the RDF representation right away, or at least got from HTML to *all* the RDF. HTH -- daniel

Reply

Daniel Kinzler

10:15 p.m.

Am 18.03.2017 um 22:48 schrieb Bob DuCharme:

New question: when I see that https://www.wikidata.org/wiki/Special:EntityData says "This page provides a linked data interface to entity values", can you tell me what "entity" means in the context of Wikidata? If I was going to refer to something that can be identified with a URI and described by triples in which it is the subject, I would just use the term "resource" as described at https://www.w3.org/TR/rdf11-concepts/#resources-and-statements (and remembering what "RDF" stands for!) so I'm guessing that "entity" means something a little more specific than that here.

The Wikidata (or technically, Wikibase) data model is not defined in terms of RDF. Have a look at the primer <https://www.mediawiki.org/wiki/Wikibase/DataModel/Primer> and the spec <https://www.mediawiki.org/wiki/Wikibase/DataModel>. Entitites are the top level elements of Wikidata. There are currently two kinds: Items (things or concepts in the world) and Properties (attributes for describing Items and other entities). Wikibase Entities are certainly Resources in the RDF sense, but so are some of the more fine grained components of the Wikibase model, such as Statements and References. You can find the OWL file for the RDF binding of Wikibase at <http://wikiba.se/ontology>. -- Daniel Kinzler Senior Software Developer Wikimedia Deutschland Gesellschaft zur Förderung Freien Wissens e.V.

Reply

Daniel Kinzler

11:16 p.m.

Am 18.03.2017 um 23:15 schrieb Daniel Kinzler:

Wikibase Entities are certainly Resources in the RDF sense, but so are some of the more fine grained components of the Wikibase model, such as Statements and References. You can find the OWL file for the RDF binding of Wikibase at <http://wikiba.se/ontology>.

If you are interested, there's a paper about mapping Wikidata to RDF: http://korrekt.org/papers/Wikidata-RDF-export-2014.pdf Note however that the mapping used on wikidata.org has somewhat diverged from what is discribed in the paper. -- Daniel Kinzler Senior Software Developer Wikimedia Deutschland Gesellschaft zur Förderung Freien Wissens e.V.

Reply

Bob DuCharme

19 Mar 19 Mar

5:21 p.m.

Thanks, that paper address just about all of the questions that were floating around in my head. I do have to ask: if the mapping used on wikidata.org has diverged from what is described there, is a more up-to-date description of the mapping available anywhere? (Thanks to Stas as well!) Bob On 03/18/2017 07:16 PM, Daniel Kinzler wrote:

Am 18.03.2017 um 23:15 schrieb Daniel Kinzler:

Wikibase Entities are certainly Resources in the RDF sense, but so are some of the more fine grained components of the Wikibase model, such as Statements and References. You can find the OWL file for the RDF binding of Wikibase at <http://wikiba.se/ontology>.

If you are interested, there's a paper about mapping Wikidata to RDF: http://korrekt.org/papers/Wikidata-RDF-export-2014.pdf Note however that the mapping used on wikidata.org has somewhat diverged from what is discribed in the paper.

Reply

Daniel Kinzler

7:57 p.m.

Am 19.03.2017 um 18:21 schrieb Bob DuCharme:

I do have to ask: if the mapping used on wikidata.org has diverged from what is described there, is a more up-to-date description of the mapping available anywhere?

The current mapping is the one described at https://www.mediawiki.org/wiki/Wikibase/Indexing/RDF_Dump_Format -- Daniel Kinzler Senior Software Developer Wikimedia Deutschland Gesellschaft zur Förderung Freien Wissens e.V.

Reply

Kingsley Idehen

27 Mar 27 Mar

1:13 p.m.

On 3/18/17 6:15 PM, Daniel Kinzler wrote:

Am 18.03.2017 um 22:48 schrieb Bob DuCharme:

New question: when I see that https://www.wikidata.org/wiki/Special:EntityData says "This page provides a linked data interface to entity values", can you tell me what "entity" means in the context of Wikidata? If I was going to refer to something that can be identified with a URI and described by triples in which it is the subject, I would just use the term "resource" as described at https://www.w3.org/TR/rdf11-concepts/#resources-and-statements (and remembering what "RDF" stands for!) so I'm guessing that "entity" means something a little more specific than that here.

The Wikidata (or technically, Wikibase) data model is not defined in terms of RDF. Have a look at the primer <https://www.mediawiki.org/wiki/Wikibase/DataModel/Primer> and the spec <https://www.mediawiki.org/wiki/Wikibase/DataModel>. Entitites are the top level elements of Wikidata. There are currently two kinds: Items (things or concepts in the world) and Properties (attributes for describing Items and other entities). Wikibase Entities are certainly Resources in the RDF sense, but so are some of the more fine grained components of the Wikibase model, such as Statements and References. You can find the OWL file for the RDF binding of Wikibase at <http://wikiba.se/ontology>.

Daniel, I see Wikidata is a collection of reified RDF Statements. I don't see how this model differs from RDF's model. It just so happens (in my eyes) that Wikidata includes description of statements about things which provides rich metadata, in line with the goals of Wikidata. -- Regards, Kingsley Idehen Founder & CEO OpenLink Software (Home Page: http://www.openlinksw.com) Weblogs (Blogs): Legacy Blog: http://www.openlinksw.com/blog/~kidehen/ Blogspot Blog: http://kidehen.blogspot.com Medium Blog: https://medium.com/@kidehen Profile Pages: Pinterest: https://www.pinterest.com/kidehen/ Quora: https://www.quora.com/profile/Kingsley-Uyi-Idehen Twitter: https://twitter.com/kidehen Google+: https://plus.google.com/+KingsleyIdehen/about LinkedIn: http://www.linkedin.com/in/kidehen Web Identities (WebID): Personal: http://kingsley.idehen.net/dataspace/person/kidehen#this : http://id.myopenlink.net/DAV/home/KingsleyUyiIdehen/Public/kingsley.ttl#this

Reply

Daniel Kinzler

2:42 p.m.

Am 27.03.2017 um 15:13 schrieb Kingsley Idehen:

I see Wikidata is a collection of reified RDF Statements. I don't see how this model differs from RDF's model. It just so happens (in my eyes) that Wikidata includes description of statements about things which provides rich metadata, in line with the goals of Wikidata.

It's a matter of perspective. I agree that Wikidata can be *represented* as a collection of reified RDF Statements. That's what we do for the query service. But I do not agree that this is what Wikidata *is*. RDF and the Wikibase model are quite different conceptually. But they are of equal power and thus formally equivalent: one can be represented using the other. Just because a Turing Machine is computationally equivalent to lambda calculus, that does not mean they are the same thing. Understanding one in terms of the other may be helpful in some context, and irrelevant in another. There is nothing special about the relationship between Wikibase/Wikidata and RDF; Wikibase has an RDF binding, but it is not defined in terms of RDF, its specification does not rely on RDF concepts. The Wikibase model can just as well (or perhaps more easily) be understood and represented in terms of the Topic Maps model (ISO 13250). Academically, the Wikibase model could perhaps be described as an extended model logic with reasoning rules for provenance. I think W. Stelzner explored related ideas in the 80s. Maybe one day I'll find the time to dig into this some more. -- Daniel Kinzler Principal Platform Engineer Wikimedia Deutschland Gesellschaft zur Förderung Freien Wissens e.V.

Reply

Kingsley Idehen

9:48 p.m.

On 3/27/17 10:42 AM, Daniel Kinzler wrote:

Am 27.03.2017 um 15:13 schrieb Kingsley Idehen:

I see Wikidata is a collection of reified RDF Statements. I don't see how this model differs from RDF's model. It just so happens (in my eyes) that Wikidata includes description of statements about things which provides rich metadata, in line with the goals of Wikidata.

It's a matter of perspective. I agree that Wikidata can be *represented* as a collection of reified RDF Statements. That's what we do for the query service. But I do not agree that this is what Wikidata *is*.

My point is that your model boils down to treating statements as "first class citizens" so to speak. If true, then it is as I described i.e., still an RDF model, but with emphasis of statement reification, which is actually a good thing.

RDF and the Wikibase model are quite different conceptually.

RDF doesn't exclude reification. Put differently, using reification doesn't amount to a new model different from that of RDF.

But they are of equal power and thus formally equivalent: one can be represented using the other. Just because a Turing Machine is computationally equivalent to lambda calculus, that does not mean they are the same thing.

I am not implying that.

Understanding one in terms of the other may be helpful in some context, and irrelevant in another. There is nothing special about the relationship between Wikibase/Wikidata and RDF; Wikibase has an RDF binding, but it is not defined in terms of RDF, its specification does not rely on RDF concepts.

RDF concepts boil down to the use of sentences to describe anything, just as we do everyday in the so-called real world.

The Wikibase model can just as well (or perhaps more easily) be understood and represented in terms of the Topic Maps model (ISO 13250). Academically, the Wikibase model could perhaps be described as an extended model logic with reasoning rules for provenance. I think W. Stelzner explored related ideas in the 80s. Maybe one day I'll find the time to dig into this some more.

I think we can just agree to disagree for now, since nothing you've stated is fundamentally contrary to my view of RDF -- as a Language for describing anything (including statements) :)

-- Regards, Kingsley Idehen Founder & CEO OpenLink Software (Home Page: http://www.openlinksw.com) Weblogs (Blogs): Legacy Blog: http://www.openlinksw.com/blog/~kidehen/ Blogspot Blog: http://kidehen.blogspot.com Medium Blog: https://medium.com/@kidehen Profile Pages: Pinterest: https://www.pinterest.com/kidehen/ Quora: https://www.quora.com/profile/Kingsley-Uyi-Idehen Twitter: https://twitter.com/kidehen Google+: https://plus.google.com/+KingsleyIdehen/about LinkedIn: http://www.linkedin.com/in/kidehen Web Identities (WebID): Personal: http://kingsley.idehen.net/dataspace/person/kidehen#this : http://id.myopenlink.net/DAV/home/KingsleyUyiIdehen/Public/kingsley.ttl#this

Reply

Daniel Kinzler

10:36 p.m.

Am 27.03.2017 um 23:48 schrieb Kingsley Idehen:

I think we can just agree to disagree for now, since nothing you've stated is fundamentally contrary to my view of RDF -- as a Language for describing anything (including statements) :)

Yes, that's what RDF is. My pint is: just because seomthing can be described in RDF doesn't mean it *is* RDF. As you said, RDF can describe anything. If anything that can be described with RDF *is* RDF, then everything is RDF. Then the term would be meaningless. The Wikibase model "is" an RDF model just as much as it "is" modal logic system, or any other sufficiently powerful formal language. -- Daniel Kinzler Senior Software Developer Wikimedia Deutschland Gesellschaft zur Förderung Freien Wissens e.V.

Reply

Markus Kroetzsch

3:42 p.m.

On 27.03.2017 15:13, Kingsley Idehen wrote:

On 3/18/17 6:15 PM, Daniel Kinzler wrote:

...

Daniel, I see Wikidata is a collection of reified RDF Statements. I don't see how this model differs from RDF's model. It just so happens (in my eyes) that Wikidata includes description of statements about things which provides rich metadata, in line with the goals of Wikidata.

Kingsley, Wikidata is not a collection of reified RDF statements, but it can partially be captured by such a collection. There are many ways of doing this; see [1] for a comparison of some of the more prominent approaches. All of these encodings can "capture" Wikidata in some way, but they are not equivalent in terms of RDF or SPARQL. It would therefore be wrong to claim that any of these possible encodings "is" Wikidata. Different RDF encodings are not only non-equivalent but also behave very differently in practice. Some queries that work well for one model are very slow or outright impossible to express in another model [1]. One can therefore not say that the encoding is just a detail and that Wikidata somehow "in principle" is RDF anyway. This said, the current Wikidata RDF export should serve the needs of most people who want to work with an RDF toolchain while having access to most Wikidata content. One cannot get all details from this projection, but one can do most practically useful things. Regards, Markus [1] Daniel Hernández, Aidan Hogan, Markus Krötzsch Reifying RDF: What Works Well With Wikidata? In Thorsten Liebig and Achille Fokoue, eds., Proceedings of the 11th International Workshop on Scalable Semantic Web Knowledge Base Systems, volume 1457 of CEUR Workshop Proceedings, 32-47, 2015. CEUR-WS.org https://iccl.inf.tu-dresden.de/web/Inproceedings3037

Reply

Kingsley Idehen

9:56 p.m.

On 3/27/17 11:42 AM, Markus Kroetzsch wrote:

On 27.03.2017 15:13, Kingsley Idehen wrote:

On 3/18/17 6:15 PM, Daniel Kinzler wrote:

...

Daniel, I see Wikidata is a collection of reified RDF Statements. I don't see how this model differs from RDF's model. It just so happens (in my eyes) that Wikidata includes description of statements about things which provides rich metadata, in line with the goals of Wikidata.

Kingsley, Wikidata is not a collection of reified RDF statements, but it can partially be captured by such a collection. There are many ways of doing this; see [1] for a comparison of some of the more prominent approaches. All of these encodings can "capture" Wikidata in some way, but they are not equivalent in terms of RDF or SPARQL. It would therefore be wrong to claim that any of these possible encodings "is" Wikidata. Different RDF encodings are not only non-equivalent but also behave very differently in practice. Some queries that work well for one model are very slow or outright impossible to express in another model [1]. One can therefore not say that the encoding is just a detail and that Wikidata somehow "in principle" is RDF anyway. This said, the current Wikidata RDF export should serve the needs of most people who want to work with an RDF toolchain while having access to most Wikidata content. One cannot get all details from this projection, but one can do most practically useful things. Regards, Markus [1] Daniel Hernández, Aidan Hogan, Markus Krötzsch Reifying RDF: What Works Well With Wikidata? In Thorsten Liebig and Achille Fokoue, eds., Proceedings of the 11th International Workshop on Scalable Semantic Web Knowledge Base Systems, volume 1457 of CEUR Workshop Proceedings, 32-47, 2015. CEUR-WS.org https://iccl.inf.tu-dresden.de/web/Inproceedings3037

Markus, Let's agree to disagree, for now :) -- Regards, Kingsley Idehen Founder & CEO OpenLink Software (Home Page: http://www.openlinksw.com) Weblogs (Blogs): Legacy Blog: http://www.openlinksw.com/blog/~kidehen/ Blogspot Blog: http://kidehen.blogspot.com Medium Blog: https://medium.com/@kidehen Profile Pages: Pinterest: https://www.pinterest.com/kidehen/ Quora: https://www.quora.com/profile/Kingsley-Uyi-Idehen Twitter: https://twitter.com/kidehen Google+: https://plus.google.com/+KingsleyIdehen/about LinkedIn: http://www.linkedin.com/in/kidehen Web Identities (WebID): Personal: http://kingsley.idehen.net/dataspace/person/kidehen#this : http://id.myopenlink.net/DAV/home/KingsleyUyiIdehen/Public/kingsley.ttl#this

Reply

Stas Malyshev

19 Mar 19 Mar

7:23 a.m.

Hi!

Could someone tell me what this means? http://wikiba.se/ontology-1.0.owl#directClaim describes it as a "Link between Wikibase Property and direct claim predicate" and the definition in https://www.mediawiki.org/wiki/Wikibase/Indexing/RDF_Dump_Format is similar. What makes a predicate a direct claim predicate?

Short addition to the great explanations by Daniel - the direct claim(s) have two properties: 1. They always have the simple values (i.e. either URL or literal or bnode), even though full values may contain more information. This means direct claims sometimes are missing secondary information, such as precision, globe, units, etc., but they are simpler to work with and enough for many simpler queries. 2. They include "best ranked" statements - i.e., if the property has preferred rank statements, then only preferred ones will be in direct claims. Otherwise, regular rank ones will be. Deprecated rank claims are never in direct claims. The preferred rank usually is used to describe "current" state of affairs - e.g. current manager of a company, major of a city, population of a country, spouse, etc. - or closest to it. If you need to see which of the full statements have the same data as direct claims, look for statements with type wikibase:BestRank. This way you can have the same data but with qualifiers, references and full values. -- Stas Malyshev smalyshev(a)wikimedia.org

Reply

2609

days inactive

2618

days old

wikidata@lists.wikimedia.org

Manage subscription

15 comments

5 participants

tags (0)

participants (5)

Bob DuCharme
Daniel Kinzler
Kingsley Idehen
Markus Kroetzsch
Stas Malyshev