Wikidata talk:WikiProject International relations

Automatic addition of signatures and ratifications of UN treaties

I wrote Java code that extracts the signatures and ratifications of a UN treaty from its UN Treaty Collection page and generates a QuickStatements batch to add them. For each signature/ratification it adds a signatory (P1891)/ratified by (P6193) statement with a point in time (P585) qualifier (the date of the signature/ratification) and a reference URL (P854) (the UNTC page) with a retrieved (P813) value (the current date). So far, I only ran this for the Biodiversity Beyond National Jurisdiction Treaty (Q48780688). Since doing this for all UN treaties will result in many thousand edits, I want to make sure I'm doing it right before I continue. This is my first major project on Wikidata; so far I’d only made a few isolated edits; so I’d be grateful for any advice. In particular:

• Is there anything else I should be adding while I’m at it? Labels, further qualifiers, ...?
• Does it make sense to add the same reference for each of these many statements? Or is there some way to add the reference once per treaty (and perhaps connect it to the statements)?
• The property ratified by (P6193) is “also known as” “ratification”, “acceptance”, “approval” and “accession”. In my view, this is unfortunate, since these are formally different processes, which are indicated separately on the UNTC pages; there should be a way to represent them separately. Since going through all the existing statements that use ratified by (P6193) to disambiguate them would be a lot of work, I was thinking of renaming this property to something like “state party” (which is the generalization of all these different ways of becoming a state party), creating new properties for the four cases (and possibly more found at UNTC) and then adding statements for both the general property and the more specific property to the UN treaties. Alternatively, the specific manner of becoming a state party could be added as a qualifier of the “state party” property. I’d appreciate your input on this since I feel I might be missing some advantages or disadvantages of these approaches, or other potential approaches.
• There’s a property participant (P710) which is sometimes used for state parties to older treaties, e.g. Sixth Russo-Turkish War (Q123070). Sometimes, state parties are listed using participant (P710) and individuals who signed the treaty document are listed using signatory (P1891), e.g. at Treaty of Amiens (Q112183). UN treaties don’t have relevant individual signers, and “participant” leaves somewhat unclear whether the state signed and/or ratified the treaty, so I figured that for modern treaties (and especially UN treaties) it would make more sense to use signatory (P1891) and ratified by (P6193) and not participant (P710) – or do you think participant (P710) should also be added (and if so, with which meaning? synonymous with “state party”, and thus with how ratified by (P6193) is currently defined?)?

Also, what would be a good way to document what I’m doing so that it can be maintained later? I was thinking of putting the Java code on GitHub and describing how to use it on my Wikidata user page – or is there a better place for something like that?

By the way, the reason I started doing this (though I think it’s also worthwhile in itself) is that I want to automate the process of displaying the signatories and parties of the treaties on a world map, and I figured that having the data for that here in Wikidata would be the best way to make this applicable to other treaties as well that don’t have UNTC pages.

Joriki (talk) 16:16, 27 November 2024 (UTC)

Hi, thanks for your efforts! With OpenRefine you can do all the changes into an item with a single edit, so that could make things a bit easier. I think you could use United Nations Treaty Collection object ID (P9966) in the references instead of reference URL (P854). But otherwise the edits look good to me. I have no opinion about changing the label of ratified by (P6193) but that it would probably be better to hold that discussion at the property's talk page. The property signatory (P1891) has been quite problematic for some time, and I would be open to creating a new signatory property to disambiguate humans and states/organizations, so that treaties could have both. Samoasambia ✎ 17:09, 27 November 2024 (UTC)

@Samoasambia Thanks, that’s very helpful!

I didn’t know about OpenRefine – I adapted the code to use that instead of QuickStatements.

I also didn’t know that a United Nations Treaty Collection object ID (P9966) links directly to a page that has all the signature and ratification information – that’s nice. The treaty itself has a statement about the UNTC object ID – should it nevertheless also be specified as a reference for each signature and ratification?

That UNTC page also has, for each ratification/acceptance/approval/accession, the date of entry into effect. (The UNTC pages I was using don’t contain that information.) This could be added as an effective date (P7588) qualifier for ratified by (P6193) – or is there a better way to include this?

I noticed in the meantime that the proposal discussion for ratified by (P6193) (here) and for a proposed property “parties” (here) referred to each other and the “parties” property wasn’t created because “this duplicates ratified by (P6193)”.

I also found that “adherence” is used as a general term for agreeing to be bound by a treaty (as distinct from the more widespread use in the sense of “compliance”) – see e.g. here, here, here, here and here This usage already has a Wikidata item: adherence (Q85315455). So we could rename the property to “adhered to by” and have a qualifier “manner of adherence”, which could take one of the existing values ratification (Q193170), accession (Q29527544) or approval (Q29526855) or the value “acceptance” that should be created. (If you don’t like “manner of adherence”, alternatives might be “manner of joining” or “manner of participation”.) I started a discussion on the property’s talk page. (Update: I modified the proposal there due to the additional considerations below.)

I like the idea of creating a new signatory property. We could call it “signer”, which seems to be far more common for signing persons than “signatory” (see e.g. this NGRAM comparison for “signer/signatory of the Constitution”). Joriki (talk) 02:14, 28 November 2024 (UTC)

@Samoasambia I just noticed a further, related problem. There doesn’t seem to be an agreed way to represent withdrawal from / denunciation of a treaty. There’s denunciation (Q6592876), but there doesn’t seem to be a corresponding property. All four cases I found treat this differently:

• International Convention for the Regulation of Whaling (Q1667878)

Here many states withdrew from the convention, some twice, one (the Netherlands) even three times. Ratification is only recorded in Wikidata alphabetically up to Benin, which includes only one of these states, Belize. There are two ratification statements for Belize, one qualified with start time (P580) and end time (P582), the other with point in time (P585). The constraints require a point in time (P585) qualifier, so there’s an issue warning on the first statement.

• Convention on preventing and combating violence against women and domestic violence (Q1697391)

Here Turkey withdrew from the convention. This is currently represented using repealed by (P2568), but that’s wrong since that property is for a document that repeals another document; accordingly, an issue warning is being displayed. (Also the statement exists twice with two dates that differ by a day, but that’s a separate mistake.)

• Kyoto Protocol (Q47359)

Here Canada withdrew from the treaty, but this isn’t represented at all; only Canada’s signature and ratification are recorded.

• Paris Agreement (Q21707860)

This currently only has single ratification statement (for China), and the withdrawal and re-acceptance of the United States isn’t recorded.

So none of the four cases of withdrawals I found is currently handled properly.

I see two different ways of approaching this. We could create a property “denounced by” with qualifier point in time (P585). In the case of multiple withdrawals and rejoinings as for the whaling treaty, there could be any number of statements with properties ratified by (P6193) and “denounced by”. An alternative would be to rename ratified by (P6193)) to “state party” with qualifiers start time (P580) and optionally end time (P582) instead of point in time (P585), and then have several statements with this property in the case of multiple withdrawals and rejoinings. I tended towards the first approach at first, but now I’m thinking that the second one actually has several advantages:

• It doesn’t require a choice about the terms for the manner of joining. (In the meantime, I found that “adherence” is sometimes used as a separate form of joining, rather than a general term for all forms of joining; see e.g. here and here.) The manners of joining could then be enumerated separately without chosing one of them as the most general form. My current favourite for the name of the qualifier that specifies the manner of joining is “form of consent” since these are all forms of giving consent to be bound by the treaty).
• It would fit well with the nominal form signatory (P1891), whereas currently there’s a slightly awkward discrepancy between the nominal signatory (P1891) and the participial ratified by (P6193) (or potentially “adhered to by”).
• It would also fit well with the Wikipedia infoboxes that often state the numbers of signatories and of parties, as do the UNTC pages (e.g. here).
• It would make it easier to count the parties. Instead of counting the states with more joinings than withdrawals, one could just count the number of “state party” statements that don’t have an end time (P582) qualifier.
• If a state is dissolved, e.g. the Socialist Federal Republic of Yugoslavia (Q83286), it’s no longer a state party. This could be indicated using the end cause (P1534) qualifier with value dissolution (Q5282797) together with the end time (P582) qualifier. I don’t see a natural way to indicate it in the “denounced by” model, and if it isn’t indicated, that would lead to a wrong count of parties.

The one major downside that I see is that this would require changing all the point in time (P585) qualifiers on existing ratified by (P6193) statements to start time (P580).

Joriki (talk) 13:59, 28 November 2024 (UTC)