Help talk:Modelling
Structure and purpose of this page
editIt is currently in early stage and it actually a mess. Several things I think we should address :
- Snipre expressed its opposition to it an showed concerns about subject overlap with Wikidata:Infobox task force. It's true that discussions probably more belongs into specific task forces, but it seems most of them are actually on property proposal pages, so I think I will give up the idea that will page serve as a discussion support. The documentation purpose still is useful and, I think, not at all in opposition with the automatic property suggestion that the development team will develop.
- The structure of the page, and the structure of the sections.
- The main purpose of this page is to document and show domain specific models, so we will need to build a notation to show those models, which was asked on molecular biology talk page. I tried to start discussions about that but not many answers except one Snipre hostile one right now. I'll try to develop a complete example with the support of TV serials (TODO, check if there is a task force for that) that will show how to use properties and items of type episode, season, series will interact, which properties and qualifier to use ... If any of you have ideas about that I'm listening :), then I will start a discussion about the utility and structure of this page with all the elements in line
- Address the misunderstanding with development team about the purpose to model the whole world is counterproductive objection. It is not the purpose of this page, nor it is to set strict rules about model conformation. TomT0m (talk) 12:39, 7 July 2013 (UTC)
Insert a note
editThis is not an official help page of wikidata, please mention that at the top of the page in order to avoid misunderstanding. If the RfC gives some outputs it will be possible to consider that as a official tool. Snipre (talk) 12:55, 6 August 2013 (UTC)
Series and sequences
editSo we have two properties part of the series (P179) and part of (P361) that can both be qualified by follows (P155) and followed by (P156). Are they redundant ? A potential difference may be:
- "part of" to mean that one element replaced another (the People's Republic of China replaced Taiwan at the United Nations)
- "series" to mean that two elements are contiguous elements of the same ordered set ('Star Wars: The Empire Strikes back' followed 'Star Wars: A New Hope').
But I am not sure it makes perfect sense, especially because the difference seems to be a difference in the meaning of "followed by" rather than in the meaning of the main property (not sure I am totally clear :\). --Zolo (talk) 15:19, 8 August 2013 (UTC)
- The meanings are close enough that they work. I think that creating different properties would create confusion. Filceolaire (talk) 23:49, 8 August 2013 (UTC)
- The two properties already exist (though we can always relaunch a deletion request).--Zolo (talk) 06:48, 9 August 2013 (UTC)
- Sorry. I meant that having two properties for follows (P155), depending on whether it was qualifying P179 or P361, would be confusing. I think P155 works in both these cases and does not need modifying. Filceolaire (talk) 21:46, 15 August 2013 (UTC)
- The two properties already exist (though we can always relaunch a deletion request).--Zolo (talk) 06:48, 9 August 2013 (UTC)
Set index articles
editOn the en-wikipedia there are w:en:Wikipedia:Set_index_articles (w:en:Category:All_set_index_articles). How to ...? --Diwas (talk) 01:16, 7 February 2015 (UTC)
- Yes, this is A problem. In Russian Wikipedia we can regard many types of disambigs as such set-index pages too. I'd propose 2 variants: 1) to regard them as topic lists, so to use is a list of (P360); 2) to regard them as some classes with similar names, so to use subclass of (P279). --Infovarius (talk) 16:26, 22 October 2015 (UTC)
- But some don't like both approaches, it seems. See e.g. this. --Infovarius (talk) 16:28, 22 October 2015 (UTC)
Modelling methodology for classes
editI would like to put more information on this page concerning modelling methodology for classes. The sort of thing I want to add is advice on how to set up classes, particularly higher-order classes.
The reason for adding some advice is that there are parts of Wikidata, including diseases and colors, where poor class modelling can be found. I think that advice on what do and what not to do would be helpful. I do realize that class modelling is a controversial topic in Wikidata, but I hope that my contributions would have a beneficial effect.
The kind of information I would add concerns the relationship between a class and its instances, the need for a firm description of what belongs in a particular class and what does not, the utility of higher-order classes, and the problems with classes that span multiple orders. Peter F. Patel-Schneider (talk) 20:33, 21 October 2015 (UTC)
- I hadn't noticed this page before; it definitely looks like it needs a lot of work - thanks for offering to help! ArthurPSmith (talk) 20:37, 21 October 2015 (UTC)
- Please feel free ! I'm the starter, I developped it for a while, but as only few Wikidatans followed, it slowly died, but if you're willing to expand it this is great news :) author TomT0m / talk page 13:21, 22 October 2015 (UTC)
Name of page
editAny particular reason this is called "Modelling" (with two 'l's)? As far as I know, "Modeling" is more common English. - Jmabel (talk) 00:12, 23 May 2018 (UTC)
- @Jmabel: See en:American_and_British_English_spelling_differences. ArthurPSmith (talk) 13:06, 23 May 2018 (UTC)
- Ah. Usually I know these, but this one I didn't. - Jmabel (talk) 15:00, 23 May 2018 (UTC)
Colors
editWe have a near-vacuous section on colors. Would anyone object if I delete that? - Jmabel (talk) 03:47, 25 May 2018 (UTC)
- I fixed this section with how I've been resolving color model(l)ing recently. Not sure if there's more to be added there? ArthurPSmith (talk) 13:29, 25 May 2018 (UTC)
Stray remark in "By domains"
editIn the section Help:Modelling#By domains we have the sentence "For most properties you can state a class of item that the property should be used with and a class of items that the property value (the item the property links to) should belong to."
- The links within this section, which just go higher up the page, seem near-useless.
- I have no idea what the remark has to do with the rest of the content of this section (which is all in subsections, although I'm about to add some at top level).
- If "you can state" this, is it supposed to be stated in some particular way in Wikidata?
If someone can flesh this out to a form that says something substantive about what to do in modelling -- I presume in modelling a property -- great. Otherwise, if no one has done that in a week or so, I plan to delete this sentence. - Jmabel (talk) 03:58, 25 May 2018 (UTC)
- I edited this to indicate how this is done with properties. However I agree it seems a bit out of place here - maybe a separate section on properties would be useful? ArthurPSmith (talk) 13:36, 25 May 2018 (UTC)
- I'll move it up into the "General" area, and start a section on "Items and properties". - Jmabel (talk) 04:04, 30 May 2018 (UTC)
Series and sequences
editI don't agree how follows (P155) and followed by (P156) are used on planets and other astronomical objects in the solar system. First of all, Solar System (Q544) is not a series, therefore no ordering can be defined on the parts and without an ordering, the concept of successor does not make sense. "List of astronomical objects in the solar system" can be sorted but not "solar system". We could now agree that these two concepts get convoluted into one item, but then still the ordering is illdefined. Astronomical objects can be sorted alphabetically, by mass, volume, density, distance from sun, distance form earth, etc. The current chosen ordering is completly arbitrary and with it the values of P155/P156. I suggest to change the example on the help page and remove on planet items all follows (P155) and followed by (P156) statements. --Pasleim (talk) 09:08, 2 June 2018 (UTC)
- Support - also it violates the constraints on part of the series (P179). You can compute the current ordering using apoapsis for example... ArthurPSmith (talk) 20:39, 3 June 2018 (UTC)
Template:Q
editThere is an overload of this template usage... So it is not working properly and displays nothing instead of label. --Infovarius (talk) 08:20, 5 June 2018 (UTC)
Prizes, awards, and recurring events generally
editHi @Jmabel: - I see you've added some notes under the Awards section. I've puzzled about this a bit and not come to a good conclusion, perhaps we can discuss it here? I'm also pinging Wikidata:WikiProject Award, though I don't know how active that is.
Notified participants of WikiProject Award
The question to me is whether the instance of (P31) relation is the right one here at all, or should all awards and prizes (other than the very specific awarding of some prize or medal to one person or organization) be considered as classes, which are subclasses via inheritance of award (Q618779)? Logically, that item claims it is "something given to a person or a group of people to recognize their excellence in a certain field", and is stated as a subclass of "bounty" or "object". That is, actual instances of "award" should be a specific monetary or other recognition given to a specific person or group. This is also related to the general issue of modeling recurring events like film festivals, sports events etc. Some awards (like Nobel prizes) occur with a set frequency; other awards (for example medals for military service) are more continuously given. Should the recurring item be considered a class with its instances the actual occurrences at specific times and places, or should it be an instance of the general type? I think there is a good argument here for considering them always to be classes.
Note this contrasts with how we treat people, organizations, buildings, locations, etc. where we do consider them individual items that may change over time but are themselves specific instances of their types. That is, the organization behind an award would be an instance of that type of organization, but the award(s) it gives would be subclass(es) of their type of award. Does this make sense to you? ArthurPSmith (talk) 13:52, 6 June 2018 (UTC)
- For what it's worth, my remarks simply reflect what I found when I looked at the items that were already linked here as examples. If the links were to bad examples, then someone should fix the content of those items, as well as the text here. I'm relatively new to Wikidata, but trying to amend this page to be actually informative to the next person trying to tread the path I'm now on. - Jmabel (talk) 15:34, 6 June 2018 (UTC)
- I think what's done right now regarding awards is pretty inconsistent/arbitrary. For example Fritz London Award (Q246592) (along with many other cases like this) is both a subclass of and an instance of award! IEEE Emanuel R. Piore Award (Q5970360) is a direct subclass of award, but also an instance of a subclass of award, which is also stated to be a metaclass, so that's odd. The value type constraint on award received (P166) allows the value to be an "instance or a subclass". So it would be nice to pick a consistent way to do this that makes sense across all of these cases. ArthurPSmith (talk) 16:11, 6 June 2018 (UTC)
- That would be great. If you can drive anything like a consensus, please document the result on this help page! - Jmabel (talk) 05:30, 7 June 2018 (UTC)
- @ArthurPSmith: Meanwhile, would it be OK with you if I quote or paraphrase that last paragraph you wrote on the help page itself, to explain to people what a mess this is right now? - Jmabel (talk) 16:04, 7 June 2018 (UTC)
- @Jmabel: You mean Help:Modelling? Yes, you can use my words that's fine! ArthurPSmith (talk) 17:44, 7 June 2018 (UTC)
- I think what's done right now regarding awards is pretty inconsistent/arbitrary. For example Fritz London Award (Q246592) (along with many other cases like this) is both a subclass of and an instance of award! IEEE Emanuel R. Piore Award (Q5970360) is a direct subclass of award, but also an instance of a subclass of award, which is also stated to be a metaclass, so that's odd. The value type constraint on award received (P166) allows the value to be an "instance or a subclass". So it would be nice to pick a consistent way to do this that makes sense across all of these cases. ArthurPSmith (talk) 16:11, 6 June 2018 (UTC)
I split this
editI ran into the limit on templates on a single page, and took the liberty of splitting this. The relevant new pages are Help:Modelling/General and Help:Modelling/By domains Help:Modelling/By domain; also, to navigate, Help:Modelling/Tabs/en. Jmabel (talk) 04:39, 8 June 2018 (UTC)
- @Jmabel: Thank you for your work on the pages! --Marsupium (talk) 13:40, 8 June 2018 (UTC)
Proposing that this no longer be considered a draft
editWikidata:Requests for comment/Help:Modelling - Jmabel (talk) 01:22, 3 July 2018 (UTC)