I guess number of instances works in more cases, for example the number of iphone ever built could work ... actually if I look at the ''Population'' article in frwiki, I get "Une population est un ensemble d'individus ou d'éléments partageant une ou plusieurs caractéristiques qui servent à les regrouper." (a population is a set of individuals or elements sharing one or several characters that allows to regroup them", this seems another word for "class" :) So I guess it's just a matter of vocabulary.

I guess the population of earth is the number of instances of the "human on earth" class :)

On the other hand, human population is kind of a natural property for a country, its number of wolfes for example may not be worth having a property. Having an item "French Wolf" with a property "number of instances" is way more convenient and generic, and could avoid the burden of having to create (potentially a big number of) properties for each kind of animals we might want to store population in some places.

2015-04-29 21:19 GMT+02:00 Markus Krötzsch <markus@semantic-mediawiki.org>:
Ok, sounds reasonable.

In all of these cases I do wonder though why we need to store the number at all. We can just count the instances, can we not? Queries allow for this (and will be an official feature in due course).

And in cases where not all instances are supposed to be on Wikidata, there are usually better ways to store the number, e.g., "population of Earth" is better than "number of instances of living person" (besides the fact that we don't have "living person").

Regards,

Markus

On 29.04.2015 20:29, Thomas Douillard wrote:
Actually, like it is, there is no /number of planets/ property, there is
a class of planet (solar system planet) together with a /number of
instances/ property. This might save us : we can have two item :
* solar system planet (old style definition) and
* solar system planet (new style definition)
  This might just put the problem in another place though :) although
this might not change the correctness of statements like /<solar
system/>  <has part> </solar system planet/ (old style)> :)


Pluto can still be an old style definition planet, and maybe <solar
system planet>
  is a subclass of </solar system planet/ (old style)>

Thinking about it, classes are naturally a good way to deal with
definitions.

2015-04-29 19:35 GMT+02:00 Markus Krötzsch
<markus@semantic-mediawiki.org <mailto:markus@semantic-mediawiki.org>>:


    Hi,

    General case first: Many statements depend on time and have an end
    date (e.g., population numbers). The general approach there is to
    (1) have a qualifier that clarifies the restricted temporal validity
    and (2) make the current statement "preferred". So your idea with
    the ranks was a good starting point, but it should be "normal" and
    "preferred" instead of "deprecated" and "normal". And infovarius was
    also right in this sense to use a temporal quantifier. Note that
    more than one statement can be preferred if more than one is current
    (this could be relevant, e.g., for the classes that Pluto is/was an
    instance of).


    However, this answer is only about the general pattern of dealing
    with things that changed over time, and the intended use of ranks in
    this case. Things might be different here. It's a special case in
    that it was not so much the world that changed but the definition,
    so the real question is what our property "number of planets" really
    means:

    (1) "Number of planets at a given time (given as a qualifier), based
    on the definition of planet adopted at this time"
    (2) "Number of planets according to the definition that was used
    when the property was introduced"
    (3) "Number of planets according to the definition that is current
    right now"

    (3) is problematic since it means that the meaning of the property
    would change over time, and statements that were true will become
    false. I would strongly discourage this. But both (1) and (2) are
    possible. If one wants to use (1) then *every* statement with this
    property must have some time qualifier -- otherwise it will not make
    any sense since one would not know which definition is meant.

    In case (2), the number of planets of our solar system is 8, and
    nothing else. It has never been 9 *according to the definition of
    planet used by this property*. So if this interpretation is adopted,
    then the statement with value 9 should really at best be there in a
    deprecated form, not in a temporal form. It could make sense to keep
    a deprecated form to warn other users that this should not be
    reintroduced.

    One could also add more options, e.g., one could have a qualifier
    that specifies the definition of planet that is used. This would be
    a bit like (1) but instead of time one would now always need to
    specify a definition, and the statements would not be temporal at
    all (the number would always remain 9 according to the old
    definition). One could still use "preferred" to mark the statement
    that is based on the most common definition.

    The world is beautifully complicated, isn't it? I'll leave it to you
    experts to discuss what makes sense here here :-)

    Best regards,

    Markus


    On 29.04.2015 18:05, Thomas Douillard wrote:

        Hi, a small question about qualifiers and ranks.

        It is well known that the number of planets changed in 2006. Or
        did it ?
        Of course, Pluto is still here, it's just its status that
        changed. The
        definition of planets changed in 2006.

        This imply that (imho), the statement  "the number of planets in the
        solar system in 9" should be deprecated. But infovarius did not
        agree
        with me and changed the rank of the claim back to normal and put
        an end
        date. I still think it should be deprecated, but it raise me a
        question:
        How are we supposed (if we are) to express an information about a
        deprecation ? Should we include something about the deprecation
        in the
        sources ? Should we have a qualifier ''deprecation date'' ?

        https://www.wikidata.org/wiki/Q17362350


        _______________________________________________
        Wikidata-l mailing list
        Wikidata-l@lists.wikimedia.org
        <mailto:Wikidata-l@lists.wikimedia.org>
        https://lists.wikimedia.org/mailman/listinfo/wikidata-l



    _______________________________________________
    Wikidata-l mailing list
    Wikidata-l@lists.wikimedia.org <mailto:Wikidata-l@lists.wikimedia.org>
    https://lists.wikimedia.org/mailman/listinfo/wikidata-l




_______________________________________________
Wikidata-l mailing list
Wikidata-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikidata-l



_______________________________________________
Wikidata-l mailing list
Wikidata-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikidata-l