Pulling Tatoeba fixes into CM

I learned yesterday that Tatoeba had 18 Hindi sentences using the Marathi word for help (muhduht) instead of the Hindi word (muhduhd). The sentence owner fixed all 18 in a jiffy, which is awesome. What’s the general process for getting these into CM? Is CM updating things on a regular basis, such that it’ll be fixed shortly? Is there some mechanism for marking a CM sentence as needing to be re-synced with Tatoeba?

2 Likes

There is a mechanism for marking a CM sentence as incorrect, and there are several mechanisms for marking a Tatoeba sentence as incorrect. There isn’t really a notion of syncing the two. Are you aware of how to report problems with sentences/translations/pronunciations at CM and at Tatoeba via CM?

I think the answer is no. I have posted on CM for a specific sentence, but that only works if there’s a moderator listening. I have posted on Tatoeba too. Luckily, there’s a very engaged native speaker who has volunteered to also fix orphaned sentences they didn’t write. So I go into Tatoeba, do a search for a unique word in the sentence, find the sentence, and then post my question along with an ‘at’ mention of this contact if the sentence isn’t owned by them. It has worked consistently on Tatoeba, so I just asked about CM.

So I’d appreciate links about reporting on CM, but still have the syncing question, so we can avoid confusing other CM learners with things that’ve been fixed on Tatoeba.

1 Like

There is a flag button that shows up once you submit the cloze word that lets you report a problem with a sentence:

After you’ve pressed the flag button, you can use the “click here” button to take you to the sentence on Tatoeba so you can report it there, and from the same dialog, you can report it at Clozemaster:

Note that this is not the same thing as clicking the “dialog box” button to post a comment about the sentence in a language-specific forum:

As to what happens behind the scenes once you report a sentence at Clozemaster, it depends on whether there’s someone handling these sentences for your language (presumably the same moderator/s who handle the sentence-specific comments you post in the forum). That didn’t happen for the first two years or so that I was playing Russian from English here, but then @mike found someone to take care of some of my requests. When that happens, you get a notification. I saw a little bit of activity, but nothing recently.

Is CM updating things on a regular basis, such that it’ll be fixed shortly?

It’s safer not to assume so.

Is there some mechanism for marking a CM sentence as needing to be re-synced with Tatoeba?

I haven’t seen any reason to suspect that there is. I simply assume that I need to report errors in both places if it’s wrong in both places, and only at Clozemaster if it has already been fixed at Tatoeba.

Glad to hear that you have good contacts for your language at Tatoeba. I do, too.

2 Likes

But what does one do when it’s been corrected on Tatoeba, but not here?

I have reported errors over and over for 6 months, but no change to anything here on the Italian course yet.

3 Likes

It would indeed be great if the corpus could somehow incorporate any changes on Tatoeba occassionally (perhaps at least yearly or so). However, I think I’ve seen sentences that were corrected on Tatoeba years ago (don’t dare estimate how long ago exactly though), that still appear in their original form here. However, the corresponding TTS files etc. for the Clozemaster Voices would also need to be updated, so I guess it might not be as straightforward as just “Hey, there’s a change there, let’s just pull it over here, and that’s that”.

I don’t think I’ve ever received any feedback on the Italian sentence reports I’ve made yet either, I guess perhaps there’s no Italian language experts at the moment to evaluate them, or we’ve just overwhelmed them, and they’re only able to process x of them every week/month?

I did report an issue with one of the Italian “Cloze-Reading” entries though, and that was resolved relatively promptly (though it didn’t require a language expert).

Here it would be great if at least indeed we could see a little flag on these sentences, if not for everyone, at least for ourselves, so the same sentence doesn’t get re-reported over and over, but I guess the sentence discussions mostly serve that purpose for us at least too (alerting ourselves and others to any issues with the sentence, or confirming whether it’s report-worthy).

1 Like

Thank you for that explanation! After many years of computer use, I rarely click on anything bright red.

This will hopefully save me some time separately going to Tatoeba.

1 Like