User talk:Metamorforme42/Ambassadors

@Metamorforme42: hello! I agree with the whole classification except the last row. I think that ambassador of France to Germany (Q20942020) should be not instance of (P31) ambassador to Germany (Q28796675) but subclass of (P279)! One can look at ambassador of France to Germany (Q20942020) as some class of human, so that e.g. we say that "Philippe Étienne (Q3381035) is a ambassador of France to Germany (Q20942020)". Sounds smooth, isn't it? Of course, at Wikidata we don't use P31 for this statement because we have position held (P39) for this. So I propose to change P31->P279 in the last row. P.S. Similar discussion is at Wikidata talk:WikiProject Heads of state and government. --Infovarius (talk) 12:32, 8 September 2017 (UTC)

@Infovarius: Hello! Does it means ambassador of France to Germany (Q20942020) is a "class of ambassador to Germany (Q28796675)"? Because this implies there are some instances of this class; and because instances of Q5 should not be instance of position, I don't see what could be instances of this class.

Also, stating P279 instead of P31 means ambassador of France to Germany (Q20942020) isn't anymore instance of position; we can add this statement directly with P31 but isn't it redundant?

I think ambassador of France to Germany (Q20942020) is (P31) a specific position ≠ (P31 a class of positions + P279 a position). Because this item can't be subclassed (give me examples if I am wrong at this point), I think saying ambassador of France to Germany (Q20942020) is (indirectly) a class of position is wrong.

I looked at Wikidata_talk:WikiProject_Heads_of_state_and_government#P31_(instance_of)_and_P279_(subclass_of), but don't fully understand it; I need to read it again. Could you please explain me what lets you think P279 is better than P31 for this specific use? — Metamorforme42 (talk) 11:28, 10 September 2017 (UTC)

Is it because you consider ambassador of France to Germany (Q20942020) as a group of humans? If it is, Jheald gave a good explanation on the project talk page, and I totally agree with this one. — Metamorforme42 (talk) 14:59, 10 September 2017 (UTC)

Yes, I consider ambassador of France to Germany (Q20942020) as a group of humans which have (had) this position. (There can be living ambassador of France to Germany (Q20942020) or deceased ambassador of France to Germany (Q20942020) - subclasses). So it is P279 Q5. But it is a position (P31), don't mix that. I disagree with Jheald's argument about this, I am prepairing contr-argument there. --Infovarius (talk) 14:12, 12 September 2017 (UTC)

@Metamorforme42: As per the discussion above, I'm pretty sure say South African Ambassador to the United States (Q1691657) should be subclass of (P279) of both ambassador to the United States of America (Q19359052) and ambassador of South Africa (Q30158578), with instance of (P31) position (Q4164871). See Mayor of London (Q38931) for an example. I've done the two ambassador positions that I found were missing stuff your way for now, but maybe consider changing that? --Reosarevok (talk) 18:04, 28 December 2017 (UTC)

Ontology for ambassadors

Hi @Metamorforme42,

I just found your userpage and am quite pleased with that, as you seem to have alot of experience with Wikidata. I am mostly busy updating Wikipedia articles on ambassadors of Germany and have recently tried to bring the wikidata entries for those persons in line. What I found when trying to run a wikidata query showing all ambassadors of Germany is a severe deficiency full of gaps and inconsistencies.

Looking at your ontology for ambassadors, I would like to raise two concerns. I will start with the fundamental one: The model to me looks overcomplicated and creates unnecessarily the need for tens (if not hundreds) of thousands of classes 'ambassador of X to Y'. In my view it would be fully sufficient to have 'ambassador' as the position and then two mandatory qualifiers: 'Sending country/organisation' and 'Host country/organisation' (country is obvious, organisation could be e.g. NATO or EU, which also send/receive ambassadors). Has such a simplified model ever been considered? What would be the right route to get such a revised model in place?

My second point would be obsolete, if the previous point is implemented. However fo now, I see an inconsistency of the 'ambassador of Germany to x' classes. About two thirds are marked as 'subclass of (P279)' 'German ambassador' and the remainder as 'is a (P31)'. My view is in line with the above discussion, i.e. tag 'ambassador of x to y' as P279 of both 'ambassador of x' and 'ambassador to Y' and classify them with 'is a' 'Position'.

With all your Wikidata experience, I would very much appreciate your help to get this area of Wikidata improved.

NB: My long term vision is to be able to create all those 'List of German ambassadors to Y' lists automatically based on correct Wikidata entries.

Regards from Hamburg Wikipeter-HH (talk) 15:19, 5 January 2023 (UTC)

Regarding question P279/P31, I would like to note that previously 90 percent of these elements followed the Metamorforme42 ontology, the old ones were maintained accordingly, and the new elements were created. About a year ago, an anon started arbitrarily, without consultation, to repair the elements one by one, and by the time they stopped, the previously unified structure had already broken down. Restoring this is obviously not possible without a bot, nor would it be lucky.

For me, the primary thing would be that the elements are uniform, it is secondary whether the data goes into P279 or P31.

See also: Wikidata_talk:WikiProject_Heads_of_state_and_government#P31_(instance_of)_and_P279_(subclass_of), Wikidata_talk:WikiProject_Heads_of_state_and_government#Current_consensus_about_ambassadors_structure. Bye! Pallor (talk) 15:43, 5 January 2023 (UTC)

Thanks for sharing the links to prior discussions.

I agree that uniformity is the primary objective and this is a fundamental problem of Wikidata seen in many other places as well, that things can be done one way or another.

The differentiation between P31 and P279 is so difficult because in everyday language the concept of class does not exist so we normally say 'Cyrill Nunn is the German ambassador to the Netherlands' and the German ambassador to the Netherlands obviously is an ambassador. In Wikidata however we reflect this as Cyrill Nunn holds the position (P39) 'German ambassador to the Netherlands', so 'German ambassador to the Netherlands' is an instance of Position. It is however not an instance of 'ambassador of Germany' nor an instance of 'ambassador to the Netherlands'. 'Ambassador of X to Y' is an abstract concept (i.e. a class) which only comes to life when a person holds that position. Therefore I plead for using P279 instead of P31.

NB: my preference still is with further simplification as mentioned in my first point above. Wikipeter-HH (talk) 14:03, 6 January 2023 (UTC)

@Wikipeter-HH: In 2017, I took a several years long wikibreak and when I ended it, every ambassador position items were in a mess (some higher class items even had their meaning reversed after some merges). I asked what I should do on Wikidata:WikiProject Heads of state and government, but eventually I gave up.

I agree with @Pallor the priority is to have unified scheme whatever it is, as long as no information is lost.

Your message makes me realize it could be a good thing to use qualifiers instead of P31/P279 to state countries (in the same fashion of what is done on Wikidata:WikiProject every politician/Political data model) and delete un-necessary classes (those creations are by the way constantly flooding my notifications), this way would end this almost unsolvable argument about P31 versus P279, and has the advantage of being flexible, very simple to understand, and to integrate better with the way other positions are handled. Are there already existing qualifiers we could use for this, or should we propose the creation of new ones? — Metamorforme42 (talk) 14:05, 6 January 2023 (UTC)

Hi @Metamorforme42,

great that you also see the potential for simplification. I am not an expert in Wikidata details, but so far I have not been able to find qualifiers which could cover the 'sending country/organisation' and 'host country/organisation' Wikipeter-HH (talk) 13:29, 7 January 2023 (UTC)

@Wikipeter-HH: You may be familiar with this one already, but I recently found property diplomatic mission sent (P531) which sort of relates to your "host country/organization" issue (as well as to my preliminary "assigned to" proposal). It doesn't seem to be exactly the same, as it specifies the embassy itself rather than the host country as its value, but I find it close enough that we should consider how these qualifiers would relate to each other. --SM5POR (talk) 13:52, 18 February 2023 (UTC)

Hi @SM5POR,

I was on vacation, so my reply is a bit delayed. I do not think that P531 is a good fit here. P531 is describing an organisational entity, rather than a role, which an ambassador is. I do prefer the solution you outlined below ("simply stack qualifiers on the main statement") in your message from 5. February.

Regards Wikipeter-HH (talk) 09:06, 27 February 2023 (UTC)

Hi @Metamorforme42,

do you have any experience with requesting new qualifiers? I am happy to help, but do not know the formal proces. Wikipeter-HH (talk) 14:22, 17 January 2023 (UTC)

Hi @Wikipeter-HH,

1. you can start by looking at Wikidata:List_of_properties if such qualifier already exist (maybe applies to jurisdiction (P1001) and located in the administrative territorial entity (P131) could be enough: this has to be investigated). If some properties are close of what you look for, but you think we cannot use them: write down they P-number somewhere, it will be useful later.
2. If this is not the case, you can search for similar proposals in Wikidata:Property_proposal/Archive (if such proposal was not successful, reading the discussion section will be useful).
3. The next step is to create a motivated property proposal (everything is explained on the page linked). It may be a good idea to include:
   1. a link to this discussion
   2. some examples of how we should use the property, including edge cases (organisations)
   3. why country (P17) is not appropriate
   4. why properties identified in step 1 are not convenient
   5. how your proposal is different from those identified at step 2

I would be really happy to answer any additional question, but I will not be very available in the next weeks, so expect delay in my responses. — Metamorforme42 (talk) 16:40, 17 January 2023 (UTC)

Hi @Metamorforme42,

have a look at Heiko Thoms. I have used qualifiers 'represents' (P1268) and 'applies to jurisdiction' (P1001). Does that make sense?

Regards Wikipeter-HH (talk) 19:01, 26 January 2023 (UTC)

Hi @Metamorforme42,

one addition: instead of P1001 we could consider to use P2541 (operating area / Wirkungsort). What do you think? Wikipeter-HH (talk) 10:18, 30 January 2023 (UTC)

Hi @Wikipeter-HH,

P2541 intended target seems to be only geographic regions/locations, what about "ambassadors of xxx to united nations" cases? Your example on Heiko Thoms seems fine to me.

What would be the impact on Embassy of Germany, Brasília (Q5369693) (especially for office held by head of the organization (P2388))? — Metamorforme42 (talk) 12:30, 30 January 2023 (UTC)

Hi @Metamorforme42,

P2388 for Embassies should be Botschafter (Q121998). Anything more is redundant and overspecified. It is pretty obvious that the head of the German embassy in Brasilia is the German ambassador to Brasil. The example actually highlights a fundamental problem of Wikidata. There is no automatism which ensures that a change of the ambassador is reflected everywhere where that item of information is used. The leader of an embassy should not be hardcoded, but be drawn from the list of persons holding that job ... Wikipeter-HH (talk) 09:15, 31 January 2023 (UTC)

@Metamorforme42 and @Wikipeter-HH: I find it great to see you attending to this area. With respect to new properties/qualifiers, I would strongly recommend you to synchronize your plans with the Wikidata:WikiProject Data Quality/Issues/P642 effort where we will probably need several new qualifiers to replace of (P642), which is destined for deprecation.

I'm right now considering two new qualifiers, 'in service of' and 'assigned to', which may be general enough to cover also your "sending/host" "country/organization" needs. While they might be used jointly for diplomats, they could also be used individually, such as Joe Biden (Q6279)position held (P39)president (Q30461)in service ofUnited States of America (Q30) and Johnnie Cochran (Q1346641)occupation (P106)defense attorney (Q46996573)assigned toO. J. Simpson murder case (Q2669947). Note that there is no need to create aggregate positions or roles such as President of the United States (Q11696) or ambassador of Germany to France (Q22145176) (unless you want to for other reasons), but you could simply stack qualifiers on the main statement, combining generic roles, corporate bodies, titles, time periods and assignments on each other (minor aggregate positions such as deputy (Q3250324) auditor (Q10949665) of the Young Pirate (Q2301613) chapter (Q67027287) of Uppsala Municipality (Q59091) may not be notable enough to have Wikidata items created for them).

We haven't written formal proposals for these properties yet, as I'd like to solicit input on how well they translate into different languages for various contexts; would it be okay to speak about being 'assigned' as a politician to a legislative committee, as a police officer to a precinct, as a judge or a lawyer to a legal case, as a security guard to a night shift, or as a sales representative to a group of customers, in German as well as Malagasy and Japanese, or do we need additional properties for specific contexts?

Please provide your comments, ideas and additional questions at Wikidata talk:WikiProject Data Quality/Issues/P642/Property labels#Seeking language advice on future qualifier proposals (and see the adjacent main page for a selection of existing properties translated into eight major languages). SM5POR (talk) 19:48, 5 February 2023 (UTC)

As for simplifying the model, I don't think it's a good idea to drop class items such as ambassador of Germany to France (Q22145176). Such classes after all are for actual concrete ambassador positions that are created, and may be filled, vacant etc. These positions are their own things, not some sort of "aggregates" as suggested above. Bunch of relevant statements can be made about these concrete positions, using appointed by (P748), organization directed by the office or position (P2389), main regulatory text (P92) etc. Often Wikipedia articles also exist for these positions, and so these concrete positions are generally notable and warrant Wikidata items anyway.

Should we simplify the model then instead I see an opportunity to simplify parent classes of these concrete positions. Namely there are pseudo-classes like ambassador to Iceland (Q30056331) and ambassador of Iceland (Q29971738) that don't really correspond to any real life entities, and which we could probably do without easily.

As for how the current use of P31/P279 came to be, I'd like to clarify that initially P279 was used to link parent classes in most items of ambassador positions, the same way as all sorts of other positions and other classes normally do. This was until 2017 when Metamorforme42 started to enforce their personal schema. To this day I haven't found or been pointed at any discussion where this schema was approved, while its obvious flaws have been pointed out in several discussions (including here at the top already back in 2017, and in other discussions/diffs referenced here). Accordingly several users (not only me) have tried to restore the correct use of P279 afterwards. In fact at some point I even was quite near at having corrected these consistently for all ambassador positions, but then unfortunately Metamorforme42, before they engaged in any new dicussion (or in previous discussions that I carefully referenced in edit summaries), switched it back to their personal schema once again. 2001:7D0:81FD:BC80:2868:62FE:48CB:D0BE 15:15, 20 May 2023 (UTC)