User talk:DavRosen

From Wikidata
Jump to: navigation, search

Previous discussion was archived at User talk:DavRosen/Flow Archive 1 on 2017-03-08.

please read Help:Description Rippitippi (talk) 11:12, 5 August 2013 (UTC)

Representation of subsets in WD like subcategories in WP[edit]

Hi! How do you think, disjoint union of (P2738)+values as qualifiers (Q23766486)+of (P642) is tool for subsets? May be there is an easier way to specify subsets (has parts of the class (P2670)?). WP has such (automatic!) tool (Wikimedia category (Q4167836), which shows a list of subcategories/subcategories+articles/articles). Pointed a superset for a set, and all its subsets are automatically displayed in the set. Very simple. But WP is human's tool for infosearch. Here still I did not find normal tool for automatic collection of subsets. What do you think about such (semi-automatic/automatic) tool? --Fractaler (talk) 07:14, 18 August 2017 (UTC)

Hi Fractaler (talkcontribslogs), I didn't make up that use of disjoint union -- I copied it from a number of other pages where I saw it used. It lists a group of subclasses (each of them has a "subclass of" pointing back here), such that each member of the present class is a member of exactly one of those subclasses -- never none of them or more than one of them. This is related only to class membership, i.e. subclasses, not just any "subsets" in a more general sense that might not be related to membership in a wikidata class. It is also unrelated to "part of" and thus "parts of the class" which both refer to what the instances are composed of (like a car "has parts of the class" wheel, while a particular instance of a car "has part" a particular instance of wheel), not their class membership (wheel is not a subclass of car!). I'm open to other ways of expressing disjoint union of subclasses of if they convey the same information (excluding the possibility of membership in none of the subclasses or more than one). There's probably someplace to discuss this with the Ontology project or others. DavRosen (talk) 17:05, 18 August 2017 (UTC)
Thanks for the answer! About set/class - I think here set = class. And if we use sets "car", "wheel", then: set "wheel" is a subset of the set "wheels" ("group of wheels"); set "group of 4 wheels" is a subset of the set "component of car"; set "component of car" is a subset of the set "components of car" ("car"). --Fractaler (talk) 15:25, 20 August 2017 (UTC)
Okay, but I'm not sure what follows from this statement about wheel vs wheels, etc. We don't have separate classes for plural and singular. We do have instance of a class (not simply a singular wheel but a specific wheel such as the Front Left Wheel of Julio's Car, if this were of interest for wikidata). I think it's okay to think of a class as a set, though arguably it's a description or specification by which we could determine whether an unknown object is an instance/member; there isn't any method by which we could name, in some sense, the members; it includes an unlimited number of "potential" members that don't (and even that can't) exist. DavRosen (talk) 15:41, 21 August 2017 (UTC)
We don't have separate classes for plural and singular - yes, but we have group (group of humans (Q16334295), of objects and so on). The group concept solves the plural problem. Instance (member) is a temporary phenomenon in WD. Now, for example, we have some instance, but in the next time it is becoming a class (set) for observer. You made instance "Front Left Wheel of Julio's Car", but then appears "Red Front Left Wheel of Julio's Car" and then "Front Left Wheel of Julio's Car" is not instance, but class. Sisyphus (Q102561)'s work. It is This Is the House That Jack Built (Q7786043)'s problem, problem of absolute path (Q37761147)/relative path (Q37761219), homonymy (Q21701659)'s problem. For unknown object we have: there are two ways of describing, or specifying the members of, a set. --Fractaler (talk) 18:17, 22 August 2017 (UTC)