Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Stas Malyshev 
Hi!

> makes sense to explicitly record the absence of a value. For example,
> one might want to say that a politician is not a member of any party,
> but one would not add this information to every human.

One common example is George Washington (Q23). While most of the US
presidents has been members of some political party, George Washington
did not like the idea of the parties and refused to join one.
Thus, https://www.wikidata.org/wiki/Q23#P102 features "no value".

> Also, I don't think that no-value statements make much sense in
> qualifiers, besides maybe serving as "documentation" for other users who

I'm just about to write a long email about it :)

-- 
Stas Malyshev
[email protected]

___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Markus Krötzsch 

On 23.04.2015 12:25, Thomas Douillard wrote:

This is a question of point of vue and how to solves conflicted
declaration, way larger than this. There could be disputes other who is
really the father of something, this would be the same.

such a statement in Wikidata means:
* This source says that this person has no child
* This (other) source says that this person has some child

If one of the source has been proven wrong, then the statement must be
marked deprecated. If it's undecided and still disputed whose right,
then we keep the two statements. This is how Wikidata was conceptually
built By Denny, Markus, Lydia and the others.

Thoams and Lydia have already explained most of it, but for the record, 
one motivation for no-value was that Freebase used to have a special 
entity that was used as a value to express that there is no value. We 
wanted to avoid this (because it easily leads to wrong query results) by 
giving the community another option for saying this directly. 
Nevertheless, novalue is used sparingly, and only in places where it 
makes sense to explicitly record the absence of a value. For example, 
one might want to say that a politician is not a member of any party, 
but one would not add this information to every human.


Also, I don't think that no-value statements make much sense in 
qualifiers, besides maybe serving as "documentation" for other users who 
see this. Otherwise, it is usually assumed that qualifiers are 
"complete", that is, what is not there can safely be assumed to not 
hold. See my female mayor query: I just check that there is no end time 
to mayorship and this works perfectly well.


In contrast, searching for all politicians that have no statement about 
their political party given would return many politicians that actually 
had a party but for which we simply did not enter it yet. No-value is 
there to help Wikidata to work for such queries, which would otherwise 
be impossible as long as Wikidata is not complete (i.e., probably 
forever ;-).


Cheers,

Markus





___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Thomas Douillard 
You should ask the people that are at this time developping it, and Lydia,
how they will handle this. Because they will, it's in the Wikibase
DataModel, the same as qualifiers.

We will probably be provided with options on how we want to handle ranks,
unknown values and so on when we will build a query in Wikibase.

2015-04-23 12:27 GMT+02:00 Gerard Meijssen :

> Hoi,
> Yes you get both results at best and maybe arbitrarily only one never mind
> which one. It requires a query engine that is so smart that it is no longer
> the kind of query engine I know exists.
> Thanks,
>  GerardM
>
>
>
> On 23 April 2015 at 12:18, Thomas Douillard 
> wrote:
>
>> You should read the helpages Gerard sometimes :)
>>
>> This is on the Wikidata plan since the beginning, it is usable since the
>> beginning, and it will be queryable ... I really don't understand what
>> bothers you.
>>
>> 2015-04-23 12:12 GMT+02:00 Gerard Meijssen :
>>
>>> Hoi,
>>> For me it is obvious ... this search for perfection is an enemy of what
>>> is good in Wikidata. It destroys the results of queries. It is not usable
>>> and it is only of use in a very small percentage of cases.
>>>
>>> Compare it with the "female mayor" question. Our issue is that we have
>>> not enough data in the first place and we should concentrate on making
>>> things easier to comprehend not weigh it down with the balast of
>>> discussions elsewhere.
>>> Thanks,
>>>   GerardM
>>>
>>> On 23 April 2015 at 11:56, Thomas Douillard 
>>> wrote:
>>>
 In wikidata, the absence of a claim about something cannot mean the
 claim has no couterpart in reality, as Wikidata is an will always be
 incomplete.

 For example if we have a series, maybe finished, maybe unfinished, we
 wil have claims that says:

 Episode 2 follows Episode 1
 Episode 3 follows Episode 2
 ...

 If for some reason we're sure that the series is other, we can state
 no value follows Episode 2

 Which means "We're sure the series is other".

 Otherwise this means "Wikidata do not know", for some reason there
 could be a following episode but noone updated Wikidata yet, for example.

 2015-04-23 11:33 GMT+02:00 Gerard Meijssen :

> Hoi,
> Sorry for being dense.. What is wrong with there being no value ?
> Having a "no value" is imho understanding only a complication of saying
> nothing... Why not say nothing in the first place ?
> Thanks,
>  GerardM
>
> On 22 April 2015 at 21:52, Markus Krötzsch <
> [email protected]> wrote:
>
>> Hi Thomas,
>>
>> On 22.04.2015 20:06, Thomas Douillard wrote:
>>
>>> Hi, there is items about Wikibase data model in Wikidata (created by
>>> me,
>>> but not only)
>>>
>>> If I understand correctly, they could be cited in the semantic web as
>>> https://www.wikidata.org/entity/Q19798647
>>>
>>
>> "No value" is exactly that: not a value. It should not be confused
>> with a (definite) value that is used with claims (as the item description
>> seems to suggest). The reason why we introduced "no value" was to be able
>> to say this without resorting to a "special value" to represent this.
>>
>> You can also find some rationale about this in our article "Wikidata:
>> a free collaborative knowledgebase" (see
>> https://ddll.inf.tu-dresden.de/web/Article4002/en). Basically, the
>> main point is that, if you are querying for two people with a common 
>> child,
>> you wouldn't want to get pairs of people who both have "novalue" as a 
>> value
>> for "child". The same is true for "some value" (sometimes referred to as
>> "unknown value") -- again, if this would be a definite "special" value, 
>> and
>> be treated like a value in queries, it would lead to wrong results.
>>
>> Cheers,
>>
>> Markus
>>
>>
>>> (If they are kept /o\)
>>>
>>> Tom²
>>>
>>>
>>> ___
>>> Wikidata-l mailing list
>>> [email protected]
>>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>>
>>>
>>
>> ___
>> Wikidata-l mailing list
>> [email protected]
>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>
>
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
>

 ___
 Wikidata-l mailing list
 [email protected]
 https://lists.wikimedia.org/mailman/listinfo/wikidata-l


>>>
>>> ___
>>> Wikidata-l mailing list
>>> [email protected]
>>> https://lists.wikimedia.org/mailman/listinfo/

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Gerard Meijssen 
Hoi,
Yes you get both results at best and maybe arbitrarily only one never mind
which one. It requires a query engine that is so smart that it is no longer
the kind of query engine I know exists.
Thanks,
 GerardM



On 23 April 2015 at 12:18, Thomas Douillard 
wrote:

> You should read the helpages Gerard sometimes :)
>
> This is on the Wikidata plan since the beginning, it is usable since the
> beginning, and it will be queryable ... I really don't understand what
> bothers you.
>
> 2015-04-23 12:12 GMT+02:00 Gerard Meijssen :
>
>> Hoi,
>> For me it is obvious ... this search for perfection is an enemy of what
>> is good in Wikidata. It destroys the results of queries. It is not usable
>> and it is only of use in a very small percentage of cases.
>>
>> Compare it with the "female mayor" question. Our issue is that we have
>> not enough data in the first place and we should concentrate on making
>> things easier to comprehend not weigh it down with the balast of
>> discussions elsewhere.
>> Thanks,
>>   GerardM
>>
>> On 23 April 2015 at 11:56, Thomas Douillard 
>> wrote:
>>
>>> In wikidata, the absence of a claim about something cannot mean the
>>> claim has no couterpart in reality, as Wikidata is an will always be
>>> incomplete.
>>>
>>> For example if we have a series, maybe finished, maybe unfinished, we
>>> wil have claims that says:
>>>
>>> Episode 2 follows Episode 1
>>> Episode 3 follows Episode 2
>>> ...
>>>
>>> If for some reason we're sure that the series is other, we can state
>>> no value follows Episode 2
>>>
>>> Which means "We're sure the series is other".
>>>
>>> Otherwise this means "Wikidata do not know", for some reason there could
>>> be a following episode but noone updated Wikidata yet, for example.
>>>
>>> 2015-04-23 11:33 GMT+02:00 Gerard Meijssen :
>>>
 Hoi,
 Sorry for being dense.. What is wrong with there being no value ?
 Having a "no value" is imho understanding only a complication of saying
 nothing... Why not say nothing in the first place ?
 Thanks,
  GerardM

 On 22 April 2015 at 21:52, Markus Krötzsch <
 [email protected]> wrote:

> Hi Thomas,
>
> On 22.04.2015 20:06, Thomas Douillard wrote:
>
>> Hi, there is items about Wikibase data model in Wikidata (created by
>> me,
>> but not only)
>>
>> If I understand correctly, they could be cited in the semantic web as
>> https://www.wikidata.org/entity/Q19798647
>>
>
> "No value" is exactly that: not a value. It should not be confused
> with a (definite) value that is used with claims (as the item description
> seems to suggest). The reason why we introduced "no value" was to be able
> to say this without resorting to a "special value" to represent this.
>
> You can also find some rationale about this in our article "Wikidata:
> a free collaborative knowledgebase" (see
> https://ddll.inf.tu-dresden.de/web/Article4002/en). Basically, the
> main point is that, if you are querying for two people with a common 
> child,
> you wouldn't want to get pairs of people who both have "novalue" as a 
> value
> for "child". The same is true for "some value" (sometimes referred to as
> "unknown value") -- again, if this would be a definite "special" value, 
> and
> be treated like a value in queries, it would lead to wrong results.
>
> Cheers,
>
> Markus
>
>
>> (If they are kept /o\)
>>
>> Tom²
>>
>>
>> ___
>> Wikidata-l mailing list
>> [email protected]
>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>
>>
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>


 ___
 Wikidata-l mailing list
 [email protected]
 https://lists.wikimedia.org/mailman/listinfo/wikidata-l


>>>
>>> ___
>>> Wikidata-l mailing list
>>> [email protected]
>>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>>
>>>
>>
>> ___
>> Wikidata-l mailing list
>> [email protected]
>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>
>>
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
>
___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Thomas Douillard 
This is a question of point of vue and how to solves conflicted
declaration, way larger than this. There could be disputes other who is
really the father of something, this would be the same.

such a statement in Wikidata means:
* This source says that this person has no child
* This (other) source says that this person has some child

If one of the source has been proven wrong, then the statement must be
marked deprecated. If it's undecided and still disputed whose right, then
we keep the two statements. This is how Wikidata was conceptually built By
Denny, Markus, Lydia and the others.

2015-04-23 12:19 GMT+02:00 Gerard Meijssen :

> Hoi,
> A question about the example... does that second source really care about
> this and what do we do when we know this to be incorrect. It is bad enough
> to want to know about the sum of all knowledge, I do not care for present
> and past mistakes.
> Thanks,
>GerardM
>
> On 23 April 2015 at 12:15, Jane Darnell  wrote:
>
>> Thanks! I added your example and Lydia's to the talk page here
>> https://www.wikidata.org/wiki/Talk:Q19798647
>>
>> On Thu, Apr 23, 2015 at 12:06 PM, Thomas Douillard <
>> [email protected]> wrote:
>>
>>> Don't use Q19798647 The item is just a description of the concept ! on
>>> Wikidata, there is a way to set an unknow value to a claim.
>>>
>>> here is how to set this special value on a claim in the UI, in french :
>>> to https://commons.wikimedia.org/wiki/File:Wikidata_pas_de_valeur.png
>>>
>>> You must click to the little blue icon on the left of the label.
>>>
>>> 2015-04-23 11:46 GMT+02:00 Jane Darnell :
>>>
 So the intention is to use Q19798647 as a random place holder
 specifically for Wikibase entities without a value, but not as a "no value"
 placeholder for a person? I think it is useful in the way you describe it,
 but I am curious how you would model the source B saying that "X had no
 child"

 On Thu, Apr 23, 2015 at 11:37 AM, Lydia Pintscher <
 [email protected]> wrote:

> On Thu, Apr 23, 2015 at 11:33 AM, Gerard Meijssen
>  wrote:
> > Hoi,
> > Sorry for being dense.. What is wrong with there being no value ?
> Having a
> > "no value" is imho understanding only a complication of saying
> nothing...
> > Why not say nothing in the first place ?
>
> It is important for cases like the following for example: Source A
> says "X had a child" and source B says "X had no child".
>
>
> Cheers
> Lydia
>
> --
> Lydia Pintscher - http://about.me/lydia.pintscher
> Product Manager for Wikidata
>
> Wikimedia Deutschland e.V.
> Tempelhofer Ufer 23-24
> 10963 Berlin
> www.wikimedia.de
>
> Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e. V.
>
> Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg
> unter der Nummer 23855 Nz. Als gemeinnützig anerkannt durch das
> Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>


 ___
 Wikidata-l mailing list
 [email protected]
 https://lists.wikimedia.org/mailman/listinfo/wikidata-l


>>>
>>> ___
>>> Wikidata-l mailing list
>>> [email protected]
>>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>>
>>>
>>
>> ___
>> Wikidata-l mailing list
>> [email protected]
>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>
>>
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
>
___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Gerard Meijssen 
Hoi,
A question about the example... does that second source really care about
this and what do we do when we know this to be incorrect. It is bad enough
to want to know about the sum of all knowledge, I do not care for present
and past mistakes.
Thanks,
   GerardM

On 23 April 2015 at 12:15, Jane Darnell  wrote:

> Thanks! I added your example and Lydia's to the talk page here
> https://www.wikidata.org/wiki/Talk:Q19798647
>
> On Thu, Apr 23, 2015 at 12:06 PM, Thomas Douillard <
> [email protected]> wrote:
>
>> Don't use Q19798647 The item is just a description of the concept ! on
>> Wikidata, there is a way to set an unknow value to a claim.
>>
>> here is how to set this special value on a claim in the UI, in french :
>> to https://commons.wikimedia.org/wiki/File:Wikidata_pas_de_valeur.png
>>
>> You must click to the little blue icon on the left of the label.
>>
>> 2015-04-23 11:46 GMT+02:00 Jane Darnell :
>>
>>> So the intention is to use Q19798647 as a random place holder
>>> specifically for Wikibase entities without a value, but not as a "no value"
>>> placeholder for a person? I think it is useful in the way you describe it,
>>> but I am curious how you would model the source B saying that "X had no
>>> child"
>>>
>>> On Thu, Apr 23, 2015 at 11:37 AM, Lydia Pintscher <
>>> [email protected]> wrote:
>>>
 On Thu, Apr 23, 2015 at 11:33 AM, Gerard Meijssen
  wrote:
 > Hoi,
 > Sorry for being dense.. What is wrong with there being no value ?
 Having a
 > "no value" is imho understanding only a complication of saying
 nothing...
 > Why not say nothing in the first place ?

 It is important for cases like the following for example: Source A
 says "X had a child" and source B says "X had no child".


 Cheers
 Lydia

 --
 Lydia Pintscher - http://about.me/lydia.pintscher
 Product Manager for Wikidata

 Wikimedia Deutschland e.V.
 Tempelhofer Ufer 23-24
 10963 Berlin
 www.wikimedia.de

 Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e. V.

 Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg
 unter der Nummer 23855 Nz. Als gemeinnützig anerkannt durch das
 Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.

 ___
 Wikidata-l mailing list
 [email protected]
 https://lists.wikimedia.org/mailman/listinfo/wikidata-l

>>>
>>>
>>> ___
>>> Wikidata-l mailing list
>>> [email protected]
>>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>>
>>>
>>
>> ___
>> Wikidata-l mailing list
>> [email protected]
>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>
>>
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
>
___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Thomas Douillard 
You should read the helpages Gerard sometimes :)

This is on the Wikidata plan since the beginning, it is usable since the
beginning, and it will be queryable ... I really don't understand what
bothers you.

2015-04-23 12:12 GMT+02:00 Gerard Meijssen :

> Hoi,
> For me it is obvious ... this search for perfection is an enemy of what is
> good in Wikidata. It destroys the results of queries. It is not usable and
> it is only of use in a very small percentage of cases.
>
> Compare it with the "female mayor" question. Our issue is that we have not
> enough data in the first place and we should concentrate on making things
> easier to comprehend not weigh it down with the balast of discussions
> elsewhere.
> Thanks,
>   GerardM
>
> On 23 April 2015 at 11:56, Thomas Douillard 
> wrote:
>
>> In wikidata, the absence of a claim about something cannot mean the claim
>> has no couterpart in reality, as Wikidata is an will always be incomplete.
>>
>> For example if we have a series, maybe finished, maybe unfinished, we wil
>> have claims that says:
>>
>> Episode 2 follows Episode 1
>> Episode 3 follows Episode 2
>> ...
>>
>> If for some reason we're sure that the series is other, we can state
>> no value follows Episode 2
>>
>> Which means "We're sure the series is other".
>>
>> Otherwise this means "Wikidata do not know", for some reason there could
>> be a following episode but noone updated Wikidata yet, for example.
>>
>> 2015-04-23 11:33 GMT+02:00 Gerard Meijssen :
>>
>>> Hoi,
>>> Sorry for being dense.. What is wrong with there being no value ? Having
>>> a "no value" is imho understanding only a complication of saying nothing...
>>> Why not say nothing in the first place ?
>>> Thanks,
>>>  GerardM
>>>
>>> On 22 April 2015 at 21:52, Markus Krötzsch <
>>> [email protected]> wrote:
>>>
 Hi Thomas,

 On 22.04.2015 20:06, Thomas Douillard wrote:

> Hi, there is items about Wikibase data model in Wikidata (created by
> me,
> but not only)
>
> If I understand correctly, they could be cited in the semantic web as
> https://www.wikidata.org/entity/Q19798647
>

 "No value" is exactly that: not a value. It should not be confused with
 a (definite) value that is used with claims (as the item description seems
 to suggest). The reason why we introduced "no value" was to be able to say
 this without resorting to a "special value" to represent this.

 You can also find some rationale about this in our article "Wikidata: a
 free collaborative knowledgebase" (see
 https://ddll.inf.tu-dresden.de/web/Article4002/en). Basically, the
 main point is that, if you are querying for two people with a common child,
 you wouldn't want to get pairs of people who both have "novalue" as a value
 for "child". The same is true for "some value" (sometimes referred to as
 "unknown value") -- again, if this would be a definite "special" value, and
 be treated like a value in queries, it would lead to wrong results.

 Cheers,

 Markus


> (If they are kept /o\)
>
> Tom²
>
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
>

 ___
 Wikidata-l mailing list
 [email protected]
 https://lists.wikimedia.org/mailman/listinfo/wikidata-l

>>>
>>>
>>> ___
>>> Wikidata-l mailing list
>>> [email protected]
>>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>>
>>>
>>
>> ___
>> Wikidata-l mailing list
>> [email protected]
>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>
>>
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
>
___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Jane Darnell 
Thanks! I added your example and Lydia's to the talk page here
https://www.wikidata.org/wiki/Talk:Q19798647

On Thu, Apr 23, 2015 at 12:06 PM, Thomas Douillard <
[email protected]> wrote:

> Don't use Q19798647 The item is just a description of the concept ! on
> Wikidata, there is a way to set an unknow value to a claim.
>
> here is how to set this special value on a claim in the UI, in french : to
> https://commons.wikimedia.org/wiki/File:Wikidata_pas_de_valeur.png
>
> You must click to the little blue icon on the left of the label.
>
> 2015-04-23 11:46 GMT+02:00 Jane Darnell :
>
>> So the intention is to use Q19798647 as a random place holder
>> specifically for Wikibase entities without a value, but not as a "no value"
>> placeholder for a person? I think it is useful in the way you describe it,
>> but I am curious how you would model the source B saying that "X had no
>> child"
>>
>> On Thu, Apr 23, 2015 at 11:37 AM, Lydia Pintscher <
>> [email protected]> wrote:
>>
>>> On Thu, Apr 23, 2015 at 11:33 AM, Gerard Meijssen
>>>  wrote:
>>> > Hoi,
>>> > Sorry for being dense.. What is wrong with there being no value ?
>>> Having a
>>> > "no value" is imho understanding only a complication of saying
>>> nothing...
>>> > Why not say nothing in the first place ?
>>>
>>> It is important for cases like the following for example: Source A
>>> says "X had a child" and source B says "X had no child".
>>>
>>>
>>> Cheers
>>> Lydia
>>>
>>> --
>>> Lydia Pintscher - http://about.me/lydia.pintscher
>>> Product Manager for Wikidata
>>>
>>> Wikimedia Deutschland e.V.
>>> Tempelhofer Ufer 23-24
>>> 10963 Berlin
>>> www.wikimedia.de
>>>
>>> Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e. V.
>>>
>>> Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg
>>> unter der Nummer 23855 Nz. Als gemeinnützig anerkannt durch das
>>> Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.
>>>
>>> ___
>>> Wikidata-l mailing list
>>> [email protected]
>>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>>
>>
>>
>> ___
>> Wikidata-l mailing list
>> [email protected]
>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>
>>
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
>
___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Gerard Meijssen 
Hoi,
For me it is obvious ... this search for perfection is an enemy of what is
good in Wikidata. It destroys the results of queries. It is not usable and
it is only of use in a very small percentage of cases.

Compare it with the "female mayor" question. Our issue is that we have not
enough data in the first place and we should concentrate on making things
easier to comprehend not weigh it down with the balast of discussions
elsewhere.
Thanks,
  GerardM

On 23 April 2015 at 11:56, Thomas Douillard 
wrote:

> In wikidata, the absence of a claim about something cannot mean the claim
> has no couterpart in reality, as Wikidata is an will always be incomplete.
>
> For example if we have a series, maybe finished, maybe unfinished, we wil
> have claims that says:
>
> Episode 2 follows Episode 1
> Episode 3 follows Episode 2
> ...
>
> If for some reason we're sure that the series is other, we can state
> no value follows Episode 2
>
> Which means "We're sure the series is other".
>
> Otherwise this means "Wikidata do not know", for some reason there could
> be a following episode but noone updated Wikidata yet, for example.
>
> 2015-04-23 11:33 GMT+02:00 Gerard Meijssen :
>
>> Hoi,
>> Sorry for being dense.. What is wrong with there being no value ? Having
>> a "no value" is imho understanding only a complication of saying nothing...
>> Why not say nothing in the first place ?
>> Thanks,
>>  GerardM
>>
>> On 22 April 2015 at 21:52, Markus Krötzsch > > wrote:
>>
>>> Hi Thomas,
>>>
>>> On 22.04.2015 20:06, Thomas Douillard wrote:
>>>
 Hi, there is items about Wikibase data model in Wikidata (created by me,
 but not only)

 If I understand correctly, they could be cited in the semantic web as
 https://www.wikidata.org/entity/Q19798647

>>>
>>> "No value" is exactly that: not a value. It should not be confused with
>>> a (definite) value that is used with claims (as the item description seems
>>> to suggest). The reason why we introduced "no value" was to be able to say
>>> this without resorting to a "special value" to represent this.
>>>
>>> You can also find some rationale about this in our article "Wikidata: a
>>> free collaborative knowledgebase" (see
>>> https://ddll.inf.tu-dresden.de/web/Article4002/en). Basically, the main
>>> point is that, if you are querying for two people with a common child, you
>>> wouldn't want to get pairs of people who both have "novalue" as a value for
>>> "child". The same is true for "some value" (sometimes referred to as
>>> "unknown value") -- again, if this would be a definite "special" value, and
>>> be treated like a value in queries, it would lead to wrong results.
>>>
>>> Cheers,
>>>
>>> Markus
>>>
>>>
 (If they are kept /o\)

 Tom²


 ___
 Wikidata-l mailing list
 [email protected]
 https://lists.wikimedia.org/mailman/listinfo/wikidata-l


>>>
>>> ___
>>> Wikidata-l mailing list
>>> [email protected]
>>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>>
>>
>>
>> ___
>> Wikidata-l mailing list
>> [email protected]
>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>
>>
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
>
___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Thomas Douillard 
Don't use Q19798647 The item is just a description of the concept ! on
Wikidata, there is a way to set an unknow value to a claim.

here is how to set this special value on a claim in the UI, in french : to
https://commons.wikimedia.org/wiki/File:Wikidata_pas_de_valeur.png

You must click to the little blue icon on the left of the label.

2015-04-23 11:46 GMT+02:00 Jane Darnell :

> So the intention is to use Q19798647 as a random place holder specifically
> for Wikibase entities without a value, but not as a "no value" placeholder
> for a person? I think it is useful in the way you describe it, but I am
> curious how you would model the source B saying that "X had no child"
>
> On Thu, Apr 23, 2015 at 11:37 AM, Lydia Pintscher <
> [email protected]> wrote:
>
>> On Thu, Apr 23, 2015 at 11:33 AM, Gerard Meijssen
>>  wrote:
>> > Hoi,
>> > Sorry for being dense.. What is wrong with there being no value ?
>> Having a
>> > "no value" is imho understanding only a complication of saying
>> nothing...
>> > Why not say nothing in the first place ?
>>
>> It is important for cases like the following for example: Source A
>> says "X had a child" and source B says "X had no child".
>>
>>
>> Cheers
>> Lydia
>>
>> --
>> Lydia Pintscher - http://about.me/lydia.pintscher
>> Product Manager for Wikidata
>>
>> Wikimedia Deutschland e.V.
>> Tempelhofer Ufer 23-24
>> 10963 Berlin
>> www.wikimedia.de
>>
>> Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e. V.
>>
>> Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg
>> unter der Nummer 23855 Nz. Als gemeinnützig anerkannt durch das
>> Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.
>>
>> ___
>> Wikidata-l mailing list
>> [email protected]
>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>
>
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
>
___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Lydia Pintscher 
Hey :)

On Thu, Apr 23, 2015 at 11:46 AM, Jane Darnell  wrote:
> I think it is useful in the way you describe it, but I am curious how you
> would model the source B saying that "X had no child"

Like this: https://www.wikidata.org/w/index.php?title=Q4115189&oldid=212444514


Cheers
Lydia

-- 
Lydia Pintscher - http://about.me/lydia.pintscher
Product Manager for Wikidata

Wikimedia Deutschland e.V.
Tempelhofer Ufer 23-24
10963 Berlin
www.wikimedia.de

Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e. V.

Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg
unter der Nummer 23855 Nz. Als gemeinnützig anerkannt durch das
Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.

___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Thomas Douillard 
In wikidata, the absence of a claim about something cannot mean the claim
has no couterpart in reality, as Wikidata is an will always be incomplete.

For example if we have a series, maybe finished, maybe unfinished, we wil
have claims that says:

Episode 2 follows Episode 1
Episode 3 follows Episode 2
...

If for some reason we're sure that the series is other, we can state
no value follows Episode 2

Which means "We're sure the series is other".

Otherwise this means "Wikidata do not know", for some reason there could be
a following episode but noone updated Wikidata yet, for example.

2015-04-23 11:33 GMT+02:00 Gerard Meijssen :

> Hoi,
> Sorry for being dense.. What is wrong with there being no value ? Having a
> "no value" is imho understanding only a complication of saying nothing...
> Why not say nothing in the first place ?
> Thanks,
>  GerardM
>
> On 22 April 2015 at 21:52, Markus Krötzsch 
> wrote:
>
>> Hi Thomas,
>>
>> On 22.04.2015 20:06, Thomas Douillard wrote:
>>
>>> Hi, there is items about Wikibase data model in Wikidata (created by me,
>>> but not only)
>>>
>>> If I understand correctly, they could be cited in the semantic web as
>>> https://www.wikidata.org/entity/Q19798647
>>>
>>
>> "No value" is exactly that: not a value. It should not be confused with a
>> (definite) value that is used with claims (as the item description seems to
>> suggest). The reason why we introduced "no value" was to be able to say
>> this without resorting to a "special value" to represent this.
>>
>> You can also find some rationale about this in our article "Wikidata: a
>> free collaborative knowledgebase" (see
>> https://ddll.inf.tu-dresden.de/web/Article4002/en). Basically, the main
>> point is that, if you are querying for two people with a common child, you
>> wouldn't want to get pairs of people who both have "novalue" as a value for
>> "child". The same is true for "some value" (sometimes referred to as
>> "unknown value") -- again, if this would be a definite "special" value, and
>> be treated like a value in queries, it would lead to wrong results.
>>
>> Cheers,
>>
>> Markus
>>
>>
>>> (If they are kept /o\)
>>>
>>> Tom²
>>>
>>>
>>> ___
>>> Wikidata-l mailing list
>>> [email protected]
>>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>>
>>>
>>
>> ___
>> Wikidata-l mailing list
>> [email protected]
>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>
>
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
>
___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Gerard Meijssen 
Hoi,
That does not clarify it. So what would we say ... it is in dispute sure
but the way we work is that we have a child as a known person. So sources
may dispute to their hearts content.. It is a relation that goes both
ways.. How do we say novalue because it has to be a person ... "John Doe"
does not apply.
Thanks,
 GerardM

On 23 April 2015 at 11:37, Lydia Pintscher 
wrote:

> On Thu, Apr 23, 2015 at 11:33 AM, Gerard Meijssen
>  wrote:
> > Hoi,
> > Sorry for being dense.. What is wrong with there being no value ? Having
> a
> > "no value" is imho understanding only a complication of saying nothing...
> > Why not say nothing in the first place ?
>
> It is important for cases like the following for example: Source A
> says "X had a child" and source B says "X had no child".
>
>
> Cheers
> Lydia
>
> --
> Lydia Pintscher - http://about.me/lydia.pintscher
> Product Manager for Wikidata
>
> Wikimedia Deutschland e.V.
> Tempelhofer Ufer 23-24
> 10963 Berlin
> www.wikimedia.de
>
> Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e. V.
>
> Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg
> unter der Nummer 23855 Nz. Als gemeinnützig anerkannt durch das
> Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Jane Darnell 
So the intention is to use Q19798647 as a random place holder specifically
for Wikibase entities without a value, but not as a "no value" placeholder
for a person? I think it is useful in the way you describe it, but I am
curious how you would model the source B saying that "X had no child"

On Thu, Apr 23, 2015 at 11:37 AM, Lydia Pintscher <
[email protected]> wrote:

> On Thu, Apr 23, 2015 at 11:33 AM, Gerard Meijssen
>  wrote:
> > Hoi,
> > Sorry for being dense.. What is wrong with there being no value ? Having
> a
> > "no value" is imho understanding only a complication of saying nothing...
> > Why not say nothing in the first place ?
>
> It is important for cases like the following for example: Source A
> says "X had a child" and source B says "X had no child".
>
>
> Cheers
> Lydia
>
> --
> Lydia Pintscher - http://about.me/lydia.pintscher
> Product Manager for Wikidata
>
> Wikimedia Deutschland e.V.
> Tempelhofer Ufer 23-24
> 10963 Berlin
> www.wikimedia.de
>
> Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e. V.
>
> Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg
> unter der Nummer 23855 Nz. Als gemeinnützig anerkannt durch das
> Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Lydia Pintscher 
On Thu, Apr 23, 2015 at 11:33 AM, Gerard Meijssen
 wrote:
> Hoi,
> Sorry for being dense.. What is wrong with there being no value ? Having a
> "no value" is imho understanding only a complication of saying nothing...
> Why not say nothing in the first place ?

It is important for cases like the following for example: Source A
says "X had a child" and source B says "X had no child".


Cheers
Lydia

-- 
Lydia Pintscher - http://about.me/lydia.pintscher
Product Manager for Wikidata

Wikimedia Deutschland e.V.
Tempelhofer Ufer 23-24
10963 Berlin
www.wikimedia.de

Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e. V.

Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg
unter der Nummer 23855 Nz. Als gemeinnützig anerkannt durch das
Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.

___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Gerard Meijssen 
Hoi,
Sorry for being dense.. What is wrong with there being no value ? Having a
"no value" is imho understanding only a complication of saying nothing...
Why not say nothing in the first place ?
Thanks,
 GerardM

On 22 April 2015 at 21:52, Markus Krötzsch 
wrote:

> Hi Thomas,
>
> On 22.04.2015 20:06, Thomas Douillard wrote:
>
>> Hi, there is items about Wikibase data model in Wikidata (created by me,
>> but not only)
>>
>> If I understand correctly, they could be cited in the semantic web as
>> https://www.wikidata.org/entity/Q19798647
>>
>
> "No value" is exactly that: not a value. It should not be confused with a
> (definite) value that is used with claims (as the item description seems to
> suggest). The reason why we introduced "no value" was to be able to say
> this without resorting to a "special value" to represent this.
>
> You can also find some rationale about this in our article "Wikidata: a
> free collaborative knowledgebase" (see
> https://ddll.inf.tu-dresden.de/web/Article4002/en). Basically, the main
> point is that, if you are querying for two people with a common child, you
> wouldn't want to get pairs of people who both have "novalue" as a value for
> "child". The same is true for "some value" (sometimes referred to as
> "unknown value") -- again, if this would be a definite "special" value, and
> be treated like a value in queries, it would lead to wrong results.
>
> Cheers,
>
> Markus
>
>
>> (If they are kept /o\)
>>
>> Tom²
>>
>>
>> ___
>> Wikidata-l mailing list
>> [email protected]
>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>
>>
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Thomas Douillard 
Hehe, we should picture this idea in help page :)
https://www.wikidata.org/entity/Q1061035
 is not a pipe.

2015-04-23 9:50 GMT+02:00 Stas Malyshev :

> Hi!
>
> > OK, I did not think it might confuse users, the label needs to be
> changed.
>
> Yes, I think it should made clear that it should not be used for actual
> values.
>
> > It's of course not intended to use in statements which have no value,
> > it's just an attempt to express wikibase data model in Wikidata.
> > https://www.wikidata.org/wiki/Q16354757 is not Wikibase Data model :)
>
> Which reminds me of https://www.wikidata.org/wiki/Q1061035 ...
>
> --
> Stas Malyshev
> [email protected]
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Stas Malyshev 
Hi!

> OK, I did not think it might confuse users, the label needs to be changed.

Yes, I think it should made clear that it should not be used for actual
values.

> It's of course not intended to use in statements which have no value,
> it's just an attempt to express wikibase data model in Wikidata.
> https://www.wikidata.org/wiki/Q16354757 is not Wikibase Data model :)

Which reminds me of https://www.wikidata.org/wiki/Q1061035 ...

-- 
Stas Malyshev
[email protected]

___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-23 Thread Thomas Douillard 
OK, I did not think it might confuse users, the label needs to be changed.

It's of course not intended to use in statements which have no value, it's
just an attempt to express wikibase data model in Wikidata.
https://www.wikidata.org/wiki/Q16354757 is not Wikibase Data model :)

I changed the label to ensure users would not confuse the items with
 the real values ...

2015-04-22 21:52 GMT+02:00 Markus Krötzsch :

> Hi Thomas,
>
> On 22.04.2015 20:06, Thomas Douillard wrote:
>
>> Hi, there is items about Wikibase data model in Wikidata (created by me,
>> but not only)
>>
>> If I understand correctly, they could be cited in the semantic web as
>> https://www.wikidata.org/entity/Q19798647
>>
>
> "No value" is exactly that: not a value. It should not be confused with a
> (definite) value that is used with claims (as the item description seems to
> suggest). The reason why we introduced "no value" was to be able to say
> this without resorting to a "special value" to represent this.
>
> You can also find some rationale about this in our article "Wikidata: a
> free collaborative knowledgebase" (see
> https://ddll.inf.tu-dresden.de/web/Article4002/en). Basically, the main
> point is that, if you are querying for two people with a common child, you
> wouldn't want to get pairs of people who both have "novalue" as a value for
> "child". The same is true for "some value" (sometimes referred to as
> "unknown value") -- again, if this would be a definite "special" value, and
> be treated like a value in queries, it would lead to wrong results.
>
> Cheers,
>
> Markus
>
>
>> (If they are kept /o\)
>>
>> Tom²
>>
>>
>> ___
>> Wikidata-l mailing list
>> [email protected]
>> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>>
>>
>
> ___
> Wikidata-l mailing list
> [email protected]
> https://lists.wikimedia.org/mailman/listinfo/wikidata-l
>
___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-22 Thread Markus Krötzsch 

Hi Thomas,

On 22.04.2015 20:06, Thomas Douillard wrote:

Hi, there is items about Wikibase data model in Wikidata (created by me,
but not only)

If I understand correctly, they could be cited in the semantic web as
https://www.wikidata.org/entity/Q19798647


"No value" is exactly that: not a value. It should not be confused with 
a (definite) value that is used with claims (as the item description 
seems to suggest). The reason why we introduced "no value" was to be 
able to say this without resorting to a "special value" to represent this.


You can also find some rationale about this in our article "Wikidata: a 
free collaborative knowledgebase" (see 
https://ddll.inf.tu-dresden.de/web/Article4002/en). Basically, the main 
point is that, if you are querying for two people with a common child, 
you wouldn't want to get pairs of people who both have "novalue" as a 
value for "child". The same is true for "some value" (sometimes referred 
to as "unknown value") -- again, if this would be a definite "special" 
value, and be treated like a value in queries, it would lead to wrong 
results.


Cheers,

Markus



(If they are kept /o\)

Tom²


___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l




___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l

Re: [Wikidata-l] items about data model concepts

2015-04-22 Thread Stas Malyshev 
Hi!

> Hi, there is items about Wikibase data model in Wikidata (created by me,
> but not only)
> 
> If I understand correctly, they could be cited in the semantic web as
> https://www.wikidata.org/entity/Q19798647

What would be the purpose of these items? I.e., what is the intended usage?

-- 
Stas Malyshev
[email protected]

___
Wikidata-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-l