search

PhonologicalCorpusTools / SLPAA

5 stars 0 forks source link

write script to convert corpora saved in pre-Oct2023 format to new location/movement tree format #234

Closed kvesik closed 7 months ago

kvesik commented 10 months ago

As it relates to issue #167 and other changes in PR #230 .

@gracemyz I know this is already on your to-do list; just wanted to make it an official github issue too. :)

kvesik commented 9 months ago

@gracemyz after our conversation at Dec 4 meeting, I wanted to take a look at your script to see if I could offer any insight on why the 2-handed signs aren't getting converted properly... but branch 234 doesn't seem to have any new content in it? Let me know where I should look :)

gracemyz commented 9 months ago

Sorry, Kaili! I thought I had pushed. It's up now.

Grace

On Wed, Dec 6, 2023, 8:17 a.m. kvesik @.***> wrote:

@gracemyz https://github.com/gracemyz after our conversation at Dec 4 meeting, I wanted to take a look at your script to see if I could offer any insight on why the 2-handed signs aren't getting converted properly... but branch 234 doesn't seem to have any new content in it? Let me know where I should look :)

— Reply to this email directly, view it on GitHub https://github.com/PhonologicalCorpusTools/SLPAA/issues/234#issuecomment-1843225540, or unsubscribe https://github.com/notifications/unsubscribe-auth/ANEPQIDOTOIIEDP4MLJ2DKTYICLCXAVCNFSM6AAAAAA6WN2XJCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNBTGIZDKNJUGA . You are receiving this because you were mentioned.Message ID: @.***>

kvesik commented 9 months ago

Thank you @gracemyz ! Also, could you describe one particular sign/module that you noticed had missing movement info after conversion so I have a specific place to start? It was the abbreviate_accident corpus that you'd made reference to at the meeting, right?

gracemyz commented 9 months ago

Yes, in the abbreviate_accident corpus, these signs don't save properly:

Although I notice that ACCIDENT v1 through v3 all save properly, and they are signs where both hands move similarly.

On Thu, Dec 7, 2023 at 9:10 AM kvesik @.***> wrote:

Thank you! Also, could you describe one particular sign/module that you noticed had missing movement info after conversion so I have a specific place to start? It was the abbreviate_accident corpus that you'd made reference to at the meeting, right?

— Reply to this email directly, view it on GitHub https://github.com/PhonologicalCorpusTools/SLPAA/issues/234#issuecomment-1845742640, or unsubscribe https://github.com/notifications/unsubscribe-auth/ANEPQIBIM4JDEBB24BJSTJTYIH2CDAVCNFSM6AAAAAA6WN2XJCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNBVG42DENRUGA . You are receiving this because you were mentioned.Message ID: @.***>

kvesik commented 9 months ago

Thanks @gracemyz ! I feel like I must be missing something... I used the ABBREVIATE_ACCIDENTv3 corpus (specifically, 2023_03_14_ABBREVIATE_ACCIDENT_v3_AA_KCH.slpaa) and did these three things:

My recorded info is in this spreadsheet-- all the necessary info seems to match in the [original corpus + frozen branch] vs [updated corpus + main branch]. Am I looking at the wrong thing, or are you and I getting different results? Or could you clarify what you mean by signs not saving properly? Thank you :)

kvesik commented 8 months ago

@gracemyz and I met on 20240108 to discuss the discrepancies we were seeing re our discussion above. It turns out that the issue is not the conversion script at all. I had x-slots turned on (manual) in my settings and Grace had them turned off, and it appears that x-slots being off was causing some of the modules to not display at all in the sign summary window (even though they are present in the underlying data structures). Made a new issue #261 for this bug.