Can't switch the expanded list of solutions option off #207
Comments
|
Thank you for your kind words about the plugin! After observing the situation for some time, it seems that the Regarding the Welsh course, do you think there is anything we can do to clean up the lists of solutions that are clearly incorrect? |
|
Thank you for the reply and putting in the effort to look into it properly.
I haven't examined the incorrect solutions in any detail but would be very
happy to put some time into doing so.
I have plenty of time but very little brain power. Before I became ill with
M.E./CFS I used to be a programmer but my brain and memory problems put a
stop to that, still if I can help compile things that you can use I would
be really nice for me to be able to be useful to someone for a little while!
Below is an example of an incorrect entry:
[Fe / Mi] goda i cyn [iddin / iddyn / idtin] nhw fynd i'r gwaith.
Coda i cyn [iddin / iddyn / idtin] nhw fynd i'r gwaith.
[Fe / Mi] wna i [godi / gody / goti] cyn [iddin / iddyn / idtin] nhw
fynd i'r gwaith.
[Gwna / wna] i [godi / gody / goti] cyn [iddin / iddyn / idtin] nhw fynd
i'r gwaith.
Bydda i'n [codi / cody / coti] cyn [iddin / iddyn / idtin] nhw fynd i'r
gwaith.
I've highlighted the invalid spellings/words. The suspect spelling look
like phonetic versions and whilst I can imagine some people in Wales using
cody and iddin when writing I very much doubt that coti or idtin would be
used (with the caveat I am infinitely surprised by what is used in the dark
distant parts of the country!). Whatever the case, these spelling do not
appear in Welsh dictionaries.
The question is how would you want to proceed? I could build a list of
correct words and spellings that are not in the dictionary, perhaps?
Let me know what works best for you - let me know if there is any way I can
help to reduce your workload and if it's within my mental capabilities then
I'll be happy to help.
…On Wed, 19 Jun 2024 at 13:04, blmage ***@***.***> wrote:
Thank you for your kind words about the plugin!
After observing the situation for some time, it seems that the
compactTranslations sent along with the session data are now always
empty, which means that we don't have a source of "canonical" solutions
anymore. I'll remove the buttons in the next release.
Regarding the Welsh course, do you think there is anything we can do to
clean up the lists of solutions that are clearly incorrect?
We already have a number of heuristics to detect incorrect words
<https://github.com/blmage/duolingo-solution-viewer/blob/master/src/common/solutions.js#L276-L366>,
so any method based on inspecting the words could help. Please let me know
if you have any specific suggestions or patterns you've noticed!
—
Reply to this email directly, view it on GitHub
<#207 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACCYOJEV4T57DBDYJEG4DPDZIFXUBAVCNFSM6AAAAABJGU674SVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNZYGUZDOMZWGI>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
|
Sorry to hear about your health issues, and thanks for offering to help nonetheless! From your example, it seems we can identify a pattern where every occurrence of For the rest, examples like the one you provided should help us identify more patterns, so please feel free to share any others you come across! |
|
Wow, that's amazing. I will go through the rest of the examples I have and
pass on the results.
By the way, I now how a spreadsheet full of Welsh sentences in the
bracketed format and I haven't found any utility to split them into
complete sentences e.g.:
To take:
Would Megan [care / mind]?
and return
Would Megan care?
Would Megan mind?
Do you happen to know of a utility that does this?
BTW, I really liked the code that I saw that was in the link you sent,
great commenting, clear and easy to understand, hats off to you!
…On Fri, 21 Jun 2024 at 11:32, blmage ***@***.***> wrote:
Sorry to hear about your health issues, and thanks for offering to help
nonetheless!
From your example, it seems we can identify a pattern where every
occurrence of di is replaced by dy and ti to generate (incorrect)
variants. After some research, it appears safe to ignore these variants, so
I'll make sure to implement this in the next release.
For the rest, examples like the one you provided should help us identify
more patterns, so please feel free to share any others you come across!
—
Reply to this email directly, view it on GitHub
<#207 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACCYOJDNJACYD5HIVCPRNWTZIP6LLAVCNFSM6AAAAABJGU674SVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCOBSGQ4DKMJYGY>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
|
Version 4.18.3 is now available, let me know what you think! Regarding your question:
None that I can think of, but knowing that this feature is more or less already built into the extension, it can probably be extracted in a quick-and-dirty form. On the same principle, I'll also consider adding an "export" option for solution lists. It should not be too complicated to implement, and may prove useful for some users. |
|
Wow, I'm blown away by your responsiveness once again! You've given
absolute top quality support and afaict you don't even offer an option for
users to donate. Do you have an Amazon wish list?
I will download the latest version and give it a test drive but I'm not
doing great at the moment and it'll probably be a day or so.
I've gathered a list of sentences that have incorrectly generated entties
and a (very) quick look confirms that most (perhaps all) conform to the
pattern of the example I sent to you.
Unfortunately, because I was concentrating on doing this, I hadn't read
your email carefully and only realised that there might be a problem with
the solution you proposed/implemented when I started to check my list
against the solution you outlined in your response of 21st June.
Your email 21st June identifies that "a pattern where every occurrence of di is
replaced by dy and ti to generate (incorrect) variants", unfortunately it
is also true that a pattern where every correct occurrence of dy is
replaced by di and ti to generate (incorrect) variants!
So, with the example I sent on the 19th June.
[Fe / Mi] wna i [godi / gody / goti] cyn [iddin / iddyn / idtin] nhw
fynd i'r gwaith.
In the first section the word godi is correct and the words gody and goti
are incorrect.
The second section fits the pattern you identified where dy is replaced
with di and dt.
I don't know what information you have about the sentences but hopefully
there is something to distinguish the generated values from the manually
entered ones which should make the solution fairly straight forward but
over to you on that one!
I'll be going through my list soon(ish) and will let you know if I come
across additional issues. I could send you the spreadhseet (Excel?) with
the examples I've come across but presume that this isn't going to be
particularly useful to you?
As regards a utlity to extract the options, were you saying that code to do
this is within your plugin code and it would therefore *just* be matter of
hacking it around a bit? If so that would be amazing, it's a bit at the
edge of my cognitive abilities nowadays and my Java skills are very rusty
but it gives me a place to start when/if I magange to get cleverer again 😂.
Thanks once again,
Peter.
…On Tue, 25 Jun 2024 at 12:29, blmage ***@***.***> wrote:
Version *4.18.3* is now available, let me know what you think!
Regarding your question:
To take:
Would Megan [care / mind]?
and return
Would Megan care?
Would Megan mind?
Do you happen to know of a utility that does this?
None that I can think of, but knowing that this feature is more or less
already built into the extension, it can probably be extracted in a
quick-and-dirty form.
On the same principle, I'll also consider adding an "export" option for
solution lists. It should not be too complicated to implement, and may
prove useful for some users.
—
Reply to this email directly, view it on GitHub
<#207 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACCYOJH42RW2NIKIK4FR4WTZJFH73AVCNFSM6AAAAABJGU674SVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCOBYGY4DGNRZG4>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
I used to have a donation button on one of my older projects, but ended up feeling better without it. The simple fact that the extension can help people is rewarding enough, but I do really appreciate your support!
Somehow the cleanup code was fine until I decided that it should apply to sorted tokens instead, because it seems that the correct words always come first in the lists. This will be fixed in the upcoming release, thanks for bringing this to my attention!
If you can find some way to highlight where the error is in each sentence, then anything will do.
Yes, the export feature is ready and will be available in the next version - I'm going to extract it to a web page where one can enter a custom list of solutions soon, probably on a GitHub page. I'll let you know when it's up! |
|
"The simple fact that the extension can help people is rewarding enough" -
I really understand that but perhaps you have a pet cause that I could
donate to, otherwise I'll pay it forward to the next person that I find in
need.
As to the generated entries issue, events have rather derailed my aim of
getting back to you quickly but I have now been through my spreadhseet.
Up until today, I had thought that any examples of ti would be wrong, but
it seems not. The following example is the worst case as it
shows valid examples of ti, di and dy in the same sentence, so without
knowing which are the generated examples...
Est [di / dy / ti] ['na / yno] [efo / gyda] [di / dy / ti] [gariad / wejen]
[␣ / di / dy / ti]?
Below are some of the more complex examples with the incorrect words marked
in red:
[Didi / Didy / Diti / Diw / Dydi / Dydy / Dyti / Dyw / Tidi / Tidy / Titi /
Tiw] Megan [ddim / ddym / dtim] yn [hoffi / licio / lico] [dillad / dyllad
/ tillad] gwyn.
Dydy Megan ddim yn hoffi dillad gwyn.
Dydy Megan ddim yn licio dillad gwyn.
Dydy Megan ddim yn lico dillad gwyn.
Dyw Megan ddim yn hoffi dillad gwyn.
Dyw Megan ddim yn licio dillad gwyn.
Dyw Megan ddim yn lico dillad gwyn.
[Ydi / Ydy / Yti] [Dilan / Dylan / Tilan] yn gwisgo [dillad / dyllad / t
illad] gwyn?
Ar hyn o bryd, [didw / dw / dydw / tidw] i [ddim / ddym / dtim] [eisiau /
isio] gadael y swyddfa.
[Dach / Dich / Dych / Tich / Ydich / Ydych / Ytich] chi [wedi / wedy /
weti] bod
mewn damwain car?
All the best,
Peter.
…On Thu, 27 Jun 2024 at 11:57, blmage ***@***.***> wrote:
You've given absolute top quality support and afaict you don't even offer
an option for users to donate. Do you have an Amazon wish list?
I used to have a donation button on one of my older projects, but ended up
feeling better without it. The simple fact that the extension can help
people is rewarding enough, but I do really appreciate your support!
In the first section the word godi is correct and the words gody and goti
are incorrect.
The second section fits the pattern you identified where dy is replaced
with di and dt.
Somehow the cleanup code was fine until I decided that it should apply to
sorted tokens instead, because it seems that the correct words always come
first in the lists. This will be fixed in the upcoming release, thanks for
bringing this to my attention!
I could send you the spreadhseet (Excel?) with the examples I've come
across but presume that this isn't going to be particularly useful to you?
If you can find some way to highlight where the error is in each sentence,
then anything will do.
Were you saying that code to do this is within your plugin code and it
would therefore *just* be matter of hacking it around a bit?
Yes, the export feature is ready and will be available in the next version
- I'm going to extract it to a web page where one can enter a custom list
of solutions soon, probably on a GitHub page. I'll let you know when it's
up!
—
Reply to this email directly, view it on GitHub
<#207 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACCYOJAVLOX4TBMQICT35QTZJPVZHAVCNFSM6AAAAABJGU674SVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCOJUGM4DGNJQG4>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
|
Sorry for the delay, I was convinced that I had posted an answer, but it turns out that I somehow did not.
Thank you for your generosity! I don't have a specific cause in mind, but paying it forward to someone else in need sounds like a great idea.
After examining various solution graphs, it seems that correct tokens always come first, so the extension now uses this fact to remove the incorrect variations. I'd be interested to know if you stumble upon a case where it likely did not work though. By the way, the solution parser is now available online: https://blmage.github.io/duo-solution-parser/, if you'd like to give it a try! |
|
<<By the way, the solution parser is now available online:
https://blmage.github.io/duo-solution-parser/, if you'd like to give it a
try!>>
That is fantastic, I can't tell you how much I appreciate it!
<<After examining various solution graphs, it seems that correct tokens
always come first, so the extension now uses this fact to remove the
incorrect variations. I'd be interested to know if you stumble upon a case
where it likely did not work though.>>
I will let you know if I find anything.
Many thanks once again,
Peter.
…On Thu, 11 Jul 2024 at 13:02, blmage ***@***.***> wrote:
Sorry for the delay, I was convinced that I had posted an answer, but it
turns out that I somehow did not.
I really understand that but perhaps you have a pet cause that I could
donate to, otherwise I'll pay it forward to the next person that I find in
need.
Thank you for your generosity! I don't have a specific cause in mind, but
paying it forward to someone else in need sounds like a great idea.
The following example is the worst case as it shows valid examples of ti,
di and dy in the same sentence.
After examining various solution graphs, it seems that correct tokens
always come first, so the extension now uses this fact to remove the
incorrect variations. I'd be interested to know if you stumble upon a case
where it likely did not work though.
By the way, the solution parser is now available online:
https://blmage.github.io/duo-solution-parser/, if you'd like to give it a
try!
—
Reply to this email directly, view it on GitHub
<#207 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACCYOJCZBZC25SYZ3INZ7QTZLZXUFAVCNFSM6AAAAABJGU674SVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEMRSG42DGNBYHA>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
A while back on my Welsh course, I clicked the button that has the fly-over hint "View the expanded list of solutions, including all automatically generated variants. More exhaustive but more readable." and I am getting lots of presumably generated garbage that includes lots of spelling that are not 'valid' Welsh.
The other option can't now be switched on and the fly over help says "[Unavailable] View the list of solutions as written by the course contributors. Less exhaustive, but more readable."
Is the currently unavailable option just no longer possible or is there something I can do to remove the unwanted versions?
Many thanks for the plugin which is brilliant!
The text was updated successfully, but these errors were encountered: