[10:34:45] good morning. [10:35:30] Good evening. What's been happening? The edit list is getting too long for me to catch up on each day. [10:35:53] are the slides from last meeting (syntactic phrase functions) available yet? [10:36:14] 'The edit list'? [10:36:26] Recentchanges (re @wmtelegram_bot: 'The edit list'?) [10:36:28] you mean recent changes? [10:36:36] * phschafft nods. [10:36:51] Not that I'm aware, I've just listened to the video (re @wmtelegram_bot: are the slides from last meeting (syntactic phrase functions) available yet?) [10:37:18] yes, seen that. But I was hopeing for a nice little PDF. [10:41:44] Today I'm wondering a bit about the middle between bottom up (words form sentences), and top down (make a sentence for this meaning) for my own project. [10:42:50] Sounds interesting (re @wmtelegram_bot: Today I'm wondering a bit about the middle between bottom up (words form sentences), and top down (make a sentence f...) [10:43:11] currently wondering if there is a way to categorise 'slots' in a sentence and then select components to form a sentence based on what slots I want to fill. [10:44:03] which slots I want to fill is something from top-down, and how I form the sentence with that is bottom-up. [10:46:49] It's linked in the newsletter (re @wmtelegram_bot: are the slides from last meeting (syntactic phrase functions) available yet?) [10:46:49] for example: {externally moved with unknown actor, from-location, to-location} -> 'the box from the attic was put into the livingroom', 'someone moved the box from the attic to the livingroom', 'the box, formally in the attic, was placed in the livingroom' [10:47:31] formally is the wrong word. [10:47:40] formerly! [10:47:51] please bear with the uneducated german! [10:50:53] Jan_ainali: found the PDF! Thank you! :) [11:04:45] It sounds like you might be thinking about “frames”. This is a good thing to think about! I hope this helps: [11:04:45] https://www.cambridge.org/core/books/pattern-construction-system/semantic-role-annotation/7A6FDD037FA3AEF48D4B23156A36615F (re @wmtelegram_bot: Today I'm wondering a bit about the middle between bottom up (words form sentences), and top down (make a sentence f...) [11:08:46] thank you. [11:11:01] I think it also comes a bit from my background: a lot of constrains based logic. so you basically put very constrain into a solver, and everything it will return is a *equally* valid solution to the problem as you provided it. [11:28:30] Sounds fair. I think linguists focus more on the inequality of valid solutions, seeking to understand why we might prefer one alternative over another, but in an NLG context, the availability of the required realizer is a critical determinant. If more than one realizer is available, we might just choose the first, but we are more likely to add an additional nice-to-have [11:28:30] constrain [11:28:32] t, such as a precedence order for the participants in the relation. (re @wmtelegram_bot: I think it also comes a bit from my background: a lot of constrains based logic. so you basically put very constrain...) [11:32:50] yes. one of the common things I do is to ensure stable order. the idea being that any stable order is better for our brains than no order at all. I guess ordering solutions is in the general area of such a nice to have constrain. [11:33:29] also, in context of generating more than one sentence one might want to mix the different solutions so the resulting text is not a list of uniform sentences. [11:35:27] for example one could keep a counter running along the text generation and select the (counter % count-of-alternatives)-th alternative. still resulting in deterministic/stable result. yet not always picking one variant over all others. [11:41:05] Yes, it’s another example of where direct realization in a single pass is sub-optimal, to whatever extent we favour either consistency or variety. (re @wmtelegram_bot: for example one could keep a counter running along the text generation and select the (counter % count-of-alternativ...) [11:48:05] * phschafft nods: however there are middle ways. such as one could pass a seed to a function. the seed is part of the article's (or calling function's) source code. hence stable. yet allowing to manually tune. [11:54:38] Yes, as we might for selecting among referring expressions for the same referent (where non-availability of an appropriate form in some language might preclude the editorially preferred option). (re @wmtelegram_bot: nods: however there are middle ways. such as one could pass a seed to a function. the seed is part of the article's ...) [11:55:43] which brings it back to a 'nice-to-have constrain'. [12:05:23] …and we’d end up realizing according to the first available realizer, given the preferences expressed up-front (re @wmtelegram_bot: which brings it back to a 'nice-to-have constrain'.) [12:11:41] in fact we do. :) [14:25:42] 9905 [14:29:39] Kanhaiya: 9905? [14:30:10] TenshiHinanawi it's the antispam bot on Telegram, it requires new entries to enter a number [14:32:31] Ah, I see. [16:39:16] Is there a difference between "run" and "execute" when talking about Wikifunctions? [16:39:28] Some WikiLambda messages say "execute" and some say "run". [16:49:59] It depends. Technically, execution should be part of evaluation, but then… so should orchestration. I would say functions (and compositions) are evaluated and implementations in code are executed. Perhaps a test case would be “run” and in its function’s evaluation some implementation would fail to execute. 🤷‍♂️ (re @amire80: Is there a difference between "run" an message> [16:50:00] d "execute" when talking about Wikifunctions?) [16:50:39] It should be clarified in the Glossary. [16:50:50] I mean, if there is really a difference. [16:50:59] If there is no difference, then perhaps the terms should be merged. [16:53:21] When is “run” used? (re @amire80: If there is no difference, then perhaps the terms should be merged.) [16:59:06] Many messages. Some examples: [16:59:06] * run Functions in the public API (user right) [16:59:08] * connect an Implementation to its Function so it can be run (user right) [16:59:09] * Run function (button) [16:59:11] * Run with latest data [16:59:12] * Run Test [16:59:14] * Unable to run function. Please try again. [16:59:15] there are many more [16:59:31] You can go to https://translatewiki.net/wiki/Special:Translate?filter=&action=page&language=he&group=ext-wikilambda and use the filter box to search for "run" [17:01:17] There's a user right called "run Functions", but its message key is "action-wikilambda-execute". Message keys are internal and mostly irrelevant for end users and even for translators, but this may be a hint that for developers, the two things are the same. [17:10:50] I suppose someone thought “evaluate” was too long 🙄 It is worth noting that “run” is hardly used in English labels, and mostly not with that sense: [17:10:50] https://www.wikifunctions.org/w/index.php?fulltext=1&ns0=1&profile=advanced&search=%3A+run&title=Special%3ASearch [17:32:38] Sardar movie by karthi [17:32:55] ? (re @Kanhaiya: Sardar movie by karthi) [17:40:57] They don’t seem to be the same to me. The request is different from its fulfilment. The rights are the right to request an evaluation or “call” a function. Connecting an implementation allows it to be selected, whether or not it can execute successfully. I think I’d only find “run” to be a natural choice when applied to a plurality. Perhaps that’s why it sounds less [17:40:57] [17:40:59] odd for tests and “Running…”. I don’t know where the English texts originate, but it’s not from the English-speaking volunteer community. (re @amire80: There's a user right called "run Functions", but its message key is "action-wikilambda-execute". Message keys are internal and m...) [17:44:36] I'm talking about translatable messages in the WikiLambda extensions. They more or less completely originate from WMF staff designers, product managers, and engineers who develop the extension. (re @Al: They don’t seem to be the same to me. The request is different from its fulfilment. The rights are the right to request an evalu...) [17:45:34] I can see how "call" is different from "execute", but I wasn't talking about "call" vs "execute". I was talking about "run" vs "execute". That distinction looks more fuzzy to me. (re @Al: They don’t seem to be the same to me. The request is different from its fulfilment. The rights are the right to request an evalu...) [17:45:51] If "run" is the same as "call", then "run" should be changed to "call". [17:46:08] If "run" includes both "call" and "execute", the Glossary needs to clarify it. [17:50:39] Let’s see what the Australian contingent thinks. I’m happy to attempt a translation into British English, which may ultimately percolate upwards from en-gb to en. (re @amire80: I can see how "call" is different from "execute", but I wasn't talking about "call" vs "execute". I was talking about "run" vs "...) [17:52:00] How is this relevant? Are there differences between the terminology of calling, executing, or running functions between American, British, and Australian English? (re @Al: Let’s see what the Australian contingent thinks. I’m happy to attempt a translation into British English, which may ultimately p...) [17:52:44] My question is what does it all _actually_ mean, so that translators to all languages would be able to translate it correctly and the end users would be able to use it correctly. [17:53:23] How would I know without consulting them? (re @amire80: How is this relevant? Are there differences between the terminology of calling, executing, or running functions between American...) [17:54:43] Well, you can ask of course, but the differences between varieties of English are not the main point. The real meaning is what the code does. [17:54:56] It actually means what WikiLambda takes it to mean, however it happens to be phrased in English. (re @amire80: My question is what does it all actually mean, so that translators to all languages would be able to translate it correctly and ...) [23:18:35] I don't have a strong opinion here. I guess run is a shorter and slightly more colloquial version of execute. I don't mind if they are changes in either direction.