Property talk:P5137

Documentation

item for this sense
concept corresponding to this sense of a lexeme. Use only on lexeme senses; do not use on items or properties

Represents

Wikidata item (Q16222597)

Data type

Item

Domain

lexeme-senses (note: this should be moved to the property statements)

Usage notes

Do not use on items or properties.

Example

no label (L2046-S1) → film (Q11424)
no label (L2521-S1) → film (Q11424)
no label (L4647-S1) → motion (Q79782)
no label (L2044-S1) → motion (Q79782)
no label (L10151-S1) → beauty (Q7242)
no label (L3360-S1) → beauty (Q7242)

How to tell not to use this property yet?

Hi,

@Tubezlob, Denny, ArthurPSmith, Deryck Chan, YULdigitalpreservation, Jura1:

This property was created a bit too early, it shouldn't be used yet but some people started using it (I just removed it from 12 lexemes). Should we put a warning in the label or the description ? some properties have (DEPRECATED) at the end or the beginning of their label, could we do something similar (but what is the antonym of deprecated?)

Cdlt, VIGNERON (talk) 05:58, 24 May 2018 (UTC)[reply]

Somehow I hadn't noticed that we would start the dictionary without definitions and senses. Sorry about that.
There are always properties that are misused and descriptions generally indicate how they should be used. This can easily be added here too.
Maybe we actually need this property that way until senses are available. Once available, the property can be adjusted.
--- Jura 06:32, 24 May 2018 (UTC)[reply]

- I think "item for this Sense" is fine. We can add [NOT READY] to the property label too. Deryck Chan (talk) 08:20, 25 May 2018 (UTC)[reply]

I don't think that's too terrible. We could, once senses are available, just move those, and let people play with it for now. --Denny (talk) 16:50, 25 May 2018 (UTC)[reply]

Please see Wikidata_talk:Lexicographical_data#Linking_items_from_lexemes.
--- Jura 10:57, 26 May 2018 (UTC)[reply]

Senses are now active and the deletion request was closed. Wikidata:Database_reports/Constraint_violations/P5137#allowed_entity_types should list uses that still need to be moved to the senses section. --- Jura 11:15, 26 October 2018 (UTC)[reply]

Denotes vs. Evokes

It seems to me that this property should be called "denotes", I have added an alias accordingly. Compare also the use of "denotes" in the ontolex standard.

However, a word sense can also evoke a concept, without denoting it: "hard" and "soft" both evoke Q46997943 (Flexibility), but neither denotes it. It would be clearer to have a separate property for that, see Wikidata:Property_proposal/Lexemes#evokes. -- Duesentrieb (talk) 11:41, 24 May 2018 (UTC)[reply]

Inverse of P6254

Hi,

For the record, see the discussion on Property talk:P6254.

Cheers, VIGNERON (talk) 17:18, 14 April 2019 (UTC)[reply]

I've removed the constraint. These two properties are not inverses. If the new one had been proposed as such I would have strongly opposed it. ArthurPSmith (talk) 20:40, 17 April 2019 (UTC)[reply]
- I am with ArthurPSmith on this issue. These properies should in my opinion not be inverses, and I am not using them as such. I am using subject lexeme (P6254) for works that discusses specific lexemes, for instance, the lexeme der (L3064) is discussed in Der - som formelt subjekt i dansk (Q58799627), De nye der-konstruktioner. En øjenvidneberetning (Q59817896) and Den danske der-konstruktion (Q59823369). — Finn Årup Nielsen (fnielsen) (talk) 08:50, 26 April 2019 (UTC)[reply]
  - I agree with all three of you. Deryck Chan (talk) 08:51, 10 May 2019 (UTC)[reply]

I concur as well. item for this sense (P5137) links from a lexeme to its meaning, but the subject of subject lexeme (P6254) are items about lexemes themselves. Also, since the constraint was removed, does anyone know why I’m still seeing it as a violation on L46038-S1? —Galaktos (talk) 09:11, 14 May 2019 (UTC)[reply]

Yeah, me too. Probably some problem with caching... --Infovarius (talk) 10:20, 15 May 2019 (UTC)[reply]

@Galaktos, Infovarius: 2 weeks later I still see the constraint, any idea why? It is still caching? it seems improbable, no? Cheers, VIGNERON (talk) 12:12, 7 June 2019 (UTC)[reply]

@VIGNERON: apparently Lea Lacroix (WMDE) created T223372 for this, but it hasn’t been fixed yet. —Galaktos (talk) 10:57, 8 June 2019 (UTC)[reply]

Use on Verb, Adjective and Adverb senses (nominalisation)

@ArthurPSmith, Micru, KaMan, Rua, VIGNERON, Jura1: This thread follows on from Wikidata:Property_proposal/nominalised_item_for_this_sense.

We need a way of linking non-noun senses (verbs, adjectives and adverbs) to items, but almost all items are labelled with nouns. Rather than create a new property, should we broaden the domain of item for this sense (P5137) to allow non-noun senses to link to the item labelled with their nominalised form? For example, should we allow: run (move quickly on foot) --(item for this sense (P5137))--> running (Q105674) ?

If so, may I update the property description of item for this sense (P5137), or add extra Wikidata property example for senses (P5977) to its property page to illustrate this?

I think that since the item is still the concept that is denoted by the verb/adjective/adverb, it should hopefully still maintain the semantics of this property. So it shouldn't be a big issue if we do this.

One important thing that will have to be done is for the bot that runs queries to auto-generate the lists of indirect synonyms and indirect translations to be updated to add a filter to the query so that it only lists those lexemes of the same lexical category. Liamjamesperritt (talk) 01:08, 13 May 2019 (UTC)[reply]

I'd be interested in designing a template or scheme for applying item for this sense (P5137) to adpositions, conjunctions andother wordswith abstractsenses,consider the English preposition "in" which has multiple senses, including:

relative position "in the house"
language or formal code of expression "book in English", "program in Algol"
future relative point in time "in five minutes"

The object items should probably be subclasses of relation (Q930933) for adpositions as they describe how the subject item relates to some other point or frame of reference, grouped in classes such as temporal, territorial, physical, contextual relations etc, and of operator (Q3354588) for conjunctions "and", "or", "if", "then" (?) and so on.

In other cases I beiieve the item for this sense (P5137) statement may be equipped with qualifierssuch as subject/object has role, thus enhancing its expressiveness several times, but I haveno specificexamples ofthat in mind right now (I could perhaps invent some).

Anybodyelse thinking along the same lines, or does my approach conflict with some already established principles regarding lexemes, senses and their items? --SM5POR (talk) 18:50, 29 January 2024 (UTC)[reply]

Examples:

Verbs would most likely point to items labelled with synonyms of the sense of the noun formed from the verb.

Ex. 1: act (do something) --(item for this sense (P5137))--> action (Q4026292)

Ex. 2: move (change position) --(item for this sense (P5137))--> motion (Q79782) ("motion" is synonymous with "movement")

Ex. 3: be (exist) --(item for this sense (P5137))--> existence (Q468777)

Ex. 4: steal (take illegally) --(item for this sense (P5137))--> theft (Q2727213) ("theft" is synonymous with "stealing")

Ex. 5: see (perceive by sight) --(item for this sense (P5137))--> visual perception (Q162668) ("visual perception" is synonymous with "seeing")

Adjectives would most likely point to items labelled with synonyms of the sense of the noun formed from the adjective.

Ex. 1: beautiful (pleasing to human senses) --(item for this sense (P5137))--> beauty (Q7242)

Ex. 2: bald (having no hair on the head) --(item for this sense (P5137))--> hair loss (Q181391)

Ex. 3: responsible (accountable) --(item for this sense (P5137))--> responsibility (Q1274115)

~~Ex. 4: high (having a specified height) --(item for this sense (P5137))--> height (Q208826)~~

Adverbs derived from adjectives "ly" would point to the same items that the corresponding adjctives would point to.

Liamjamesperritt (talk) 07:16, 14 May 2019 (UTC)[reply]

I don't think a separate item is needed for nominalized items. Items aren't nouns, they're not actually any lexical category, they're concepts, so equating them with nouns is incorrect. That we commonly describe/label them with nouns is irrelevant. The verb "walk" and the noun "walking" refer to the same concept, lexical category doesn't matter. Using different properties creates an unnecessary distinction between nouns and non-nouns and falsely equates items with nouns. —Rua (mew) 11:27, 13 May 2019 (UTC)[reply]

@Rua: Thanks for your comment. Just to clarify, I didn't mean to create a separate item when I said "nominalised item". I just meant, would it be OK to use item for this sense (P5137) to link a verb sense (such as "walk") to the item labeled with its nominalised form (i.e. "walking"). So it sounds like we agree that that should be OK. But your comment was very well put, and it makes a lot of sense. I was mistakenly thinking of items as noun senses, rather than concepts independent of lexical category. Liamjamesperritt (talk) 12:03, 13 May 2019 (UTC)[reply]

It might be worth looking at a few samples before starting to link anything vaguely related with this properties. --- Jura 04:45, 14 May 2019 (UTC)[reply]

@Jura1: I added a few examples above of how I see it potentially being used. Liamjamesperritt (talk) 07:16, 14 May 2019 (UTC)[reply]

Now this property is used for getting indirect synonims and indirect translations. I don't think that a verb and a noun could be synonims. --Infovarius (talk) 15:52, 13 May 2019 (UTC)[reply]

@Infovarius: I agree, however it was argued in Wikidata:Property_proposal/nominalised_item_for_this_sense that the query used for generating indirect synonyms could just have a "lexical category filter" built into the query (so that only nouns are grouped with nouns, and verbs with verbs). And that is definitely possible to do. But that query would need to be changed. Liamjamesperritt (talk) 00:05, 14 May 2019 (UTC)[reply]

Support I would support using for « Verb, Adjective and Adverb senses », and even further I would support using this property for each and any Lexemes. Despite multiple discussions, I still don't see reason not to, it create no problems and it far better and easier to have only one property for the same thing. Cheers, VIGNERON (talk) 12:59, 14 May 2019 (UTC)[reply]
Support Ok with me. ArthurPSmith (talk) 18:29, 22 May 2019 (UTC)[reply]
@Rua, Jura1: Any opinions on this given the updated proposal? Liamjamesperritt (talk) 08:45, 30 May 2019 (UTC)[reply]
- @Liamjamesperritt: I'm not quite sure what to write. It makes sense for the samples given above when looking at them with item labels in English and it could probably work well in a single editor environment. Obviously it's still possible that for some reason there is an item with the label "beautiful" and this gets linked instead.
  The additional question here is how it would scale to Wikidata's multi-lingual, multi-editor approach. Items have gotten better since the initial import, but it's still possible that not everyone gets the same definition for an item (depending on label/description in a specific language, maybe a sitelink).
  Maybe you have tried pltools/recentdeaths. If you click on "get full article", words and expressions around death get highlighted in red. These are either labels or aliases in a series of languages of items for related concepts. I found it somewhat tricky to identify, cleanup and define aliases for some of the items. Some words could obviously appear on several items. BTW, if you find words in the articles that could be useful, but aren't highlighted, please add them to a relevant item. Maybe it gets picked-up on the next update. --- Jura 13:16, 1 June 2019 (UTC)[reply]
- @Jura1: The problem is, the community does not like the idea of adding items that have adjectives or verbs as labels (e.g. beautiful). If that were an option, then this proposal would be unnecessary. But that's not the case. However, as explained by Rua, items aren't word senses; they are concepts independent of lexical category. So the concept denoted by "beauty" is the same as the concept denoted by "beautiful". On the subject of other languages, would you like me to provide examples of how this would be done in a variety of languages? P.S. I haven't looked at pltools/recentdeaths yet, but I'll take a look when I get a chance. Best, Liamjamesperritt (talk) 01:58, 22 June 2019 (UTC)[reply]
  - Not always concept is independent of lexical category. Consider adverbs. "More beautiful" (if you don't like English, in Russian we have one word for this) doesn't mean "beauty" at all - it can be very ugly but (slightly) less ugly than a neighbour. Also I don't think that we have items suited for verbs. --Infovarius (talk) 21:34, 23 June 2019 (UTC)[reply]
  - @Infovarius: Even though "More beautiful" does not imply "Full of beauty", it is still semantically capturing the concept of "some degree of beauty". Whether it be a high or low degree of beauty, the abstract concept is still "Beauty". Regardless, "More beautiful" is just the comparative form of "Beautiful". Forms do not have their own sense. Senses only apply to the root lexeme. Liamjamesperritt (talk) 12:28, 2 July 2019 (UTC)[reply]

Support I see a potential problem for verbs. There could be an ambiguity that conflates in a single Wikidata item the action and the result of the action, for example « the process of running » for the verb « to run », and the result of an action, for example, a race. In math, it’s seem that we conflate a transformation with the result of the transformation of the same item (for example the « fourier transform » or « fourier transformation » ? « The sum » (the result of an addition) and the addition itself … It’s pretty clear that « to add » correspond to the process of adding numbers, and we have items for this, however. This is not always the case, but the best solution could be to create the items if they are missing, I think we have properties to link process to their results

I don’t think it’s easy to resolve the ambiguities without creating new items. If we take an example that should be easy, say « to run », it turns out it’s a mess. The frwiki article we are redirected to on Wikipedia if we search for « courir » is fr:Course_à_pied, whose first sentence is

« La course à pied est, avec la marche, l'un des deux modes de locomotion bipèdes de l'être humain.

, the corresponding Wikidata item is running (Q105674)  

whose instance of (P31) is « sport discipline » … and subclass of (P279) locomotion mode (probably an incorrect use of « subclass of » here, but that’s not the point). There is so many problems in this. The situation is less than clear about the meaning of this item. To me running for sport is very different from running for one life or to feed oneself … This kind of ambiguity could lead to problems like several senses pointing to the same items, but if it can lead commmunity to become more conscious of this it’s a good thing.

I think being reluctant to create items for the actions referred to verbs could lead to ambiguities in some cases, on the other hand. But it’s not a reason to oppose, if done correctly this is a good idea. Not everything is done ideally on Wikidata however. author TomT0m / talk page 09:02, 29 October 2019 (UTC)[reply]

Since there seems to be some general support, I'll go ahead and update the usage examples on the item for this sense (P5137) property page to make the increase in scope explicit. Liamjamesperritt (talk) 01:06, 30 October 2019 (UTC)[reply]

@ArthurPSmith, Micru, KaMan, Rua, VIGNERON, Jura1: @TomT0m, Liamjamesperritt, Infovarius: Sorry for not discovering and joining this discussion before. I am not really sure that mixing up the lexical categories is a good idea. It is not clear to me whether we are not degrading the semantics of the Wikidata lexemes if we merge the categories. Here are some specific thoughts:

I have always thought of the Q-items as on the same level as wordnet synsets. Synsets are as far as I can tell not merging lexical categories. We already have a few properties for wordnets (babelnet ID and ILI ID). There would no longer be a one-to-one relationship between wordnet synsets and Q-items if we start using P5137 for non-nouns.
Verbs can related to nouns in many ways, e.g., employ (L5510) could go to employer (Q3053337), employee (Q703534) or employment (Q656365) (employer, employee, employment). (This may be in line with TomT0m's comment). I am also not convinced with the "walk"/"walking" example above. "Walker" or "pedestrian" (pedestrian (Q221488)) is another noun/concept that relates to walk. And walking (Q6537379) and pedestrian (Q221488) are not synonyms.
In the example above, I am unsure why one of the senses of high should be linked to height (Q208826): How would you then linked a sense of "low"?
I have not looked deep into wordnets' verb-noun relationships. The only one I have found so far is "derivationally related" which corresponds to our derived from lexeme (P5191) I suppose.
I am wondering whether it wouldn't be better to create new properties that describe the relationship from non-nouns to Q-items, e.g., employ -> verb with agent (nomen agentis) -> employer (Q3053337), employ -> verb with patient (nomen patientis) -> employee (Q703534).

I am unsure whether we can get any help from linguistics. I have just noted the concept of lexical function (Q17163176). — Finn Årup Nielsen (fnielsen) (talk) 18:50, 19 December 2019 (UTC)[reply]

As far as I can see we usually have valency (Q1412952) 1 different concepts for a verb. For instance, verbs associated with the act of transfering possession one would have 4, e.g., "donate": donor, donee/receiver/benefactor, donated object (present/gift), donation. — Finn Årup Nielsen (fnielsen) (talk) 19:00, 19 December 2019 (UTC)[reply]

@Fnielsen: very good points. Do you have any statistics on the relative number of concepts for verbs & adjectives vs (common) nouns? It seems from what we've added so far they are likely to be a relatively small fraction. If there are 10,000 of them, that's really not a big dent in what Wikidata has... In general I think now that it might be better for us to just push to create these non-noun concept items. ArthurPSmith (talk) 19:13, 19 December 2019 (UTC)[reply]

@Fnielsen: long-waited and reasonable critics! @ArthurPSmith: each (common) noun can have corresponding adjective and several corresponding actions (verbs). But how many common nouns?.. --Infovarius (talk) 23:02, 20 December 2019 (UTC)[reply]

@Infovarius: "Each"? What would be the adjectives or verbs associated with common nouns like "carrot"? "sparrow"? "mansion"? Perhaps you are thinking only of abstract nouns? For common nouns there are likely as many as there are "class" items in Wikidata - hundreds of thousands at least. ArthurPSmith (talk) 02:48, 21 December 2019 (UTC)[reply]

@Fnielsen, Infovarius, ArthurPSmith: It seems to me that if we were to link verbs to noun Items, then it would only make sense to link them to abstract nouns in most cases, as demonstrated by the examples above (I have removed the 'height' example, as I agree that was stretching the usage a little too far). For the 'employ' case, the only abstract noun that would work would be 'employment' , as 'employment' is the only noun that represents precisely the same concept as 'to employ'. Liamjamesperritt (talk) 02:29, 28 December 2019 (UTC)[reply]

@ArthurPSmith: carrot-> "морковный" (i.e. in "carrot juice"), "sparrow" -> "воробьиный" and so on. Most concrete nouns has corresponding adjective in Russian (and even several), I ain't sure about abstract. --Infovarius (talk) 21:12, 30 December 2019 (UTC)[reply]

@Infovarius: I’m wondering, are those actually adjectives in the same sense as English adjectives, in particular, can they be used both attributively and predicatively? For example, can one say something like Сок есть морковный
(not that I know any Russian)?―BlaueBlüte (talk) 06:16, 31 December 2019 (UTC)[reply]

@BlaueBlüte: Hm, interesting question. The sentence Сок есть морковный
is understandable but doesn't look like correct. Variant Сок является морковным
is another direct translation of "Juice is carrot" (this predicative use did you mean?), it is very formal but gramatically correct. Even more predicative use is when using short form of adjectives (but they exist only for qualitative adjectives) like Сок красен
. --Infovarius (talk) 15:21, 1 January 2020 (UTC)[reply]

All this goes heavily into verb semantics. The most comprehensive and systematic (as opposed to ‘human instinct-derived’) discussion of verb semantics that I'm aware of is the lexical semantics of Latejami, by Richard A. Morneau (http://www.rickmor.x10.mx/lexical_semantics.html). He discusses the Agent/Patient/Focus stative/dynamic building blocks of verb semantics, and I think without understanding that complexity, it would be difficult to add verb-y concepts to Wikidata. -- 185.132.19.30 11:53, 25 October 2021 (UTC)[reply]

Question What’s the origin of this notion that translations need to be of the same lexical category? That would seem to me very much like a translator’s rookie mistake. Cf.

English	I	am	in	a lot noun (Q1084)	of	debt. noun (Q1084)
German	Ich	bin		hoch adverb (Q380057)		verschuldet. adjective (Q34698)
item for the respective sense				“high magnitude”		debt (Q3196867)

and many more examples. Pinging @Liamjamesperritt and @fnielsen.―BlaueBlüte (talk) 05:54, 31 December 2019 (UTC)[reply]

Should every sense ideally have an item attached to it?

I'm talking about everyday things like dishwater (Q111164108). It is a definable concept that exists in some languages --Loominade (talk) 07:39, 18 March 2022 (UTC)[reply]

I think every sense deserves at least some item but it doesn'thave to be unique, in case "dishwater" wouldn't already have had an item of its own, I would suggest linking the sense to water (Q283) with one or two qualifiers added. The same for other compound words like "wastewater" and "freshwater" --SM5POR (talk) 19:03, 29 January 2024 (UTC)[reply]

Property talk:P5137

Contents

Documentation

How to tell not to use this property yet?

Denotes vs. Evokes

Inverse of P6254

Use on Verb, Adjective and Adverb senses (nominalisation)

Should every sense ideally have an item attached to it?

Navigation menu

Property talk:P5137

Documentation

How to tell not to use this property yet?

Denotes vs. Evokes

Inverse of P6254

Use on Verb, Adjective and Adverb senses (nominalisation)

Should every sense ideally have an item attached to it?

Navigation menu

Search