search

Signbank / Global-signbank

An online sign dictionary and sign database management system for research purposes. Developed originally by Steve Cassidy/ This repo is a fork for the Dutch version, previously called 'NGT-Signbank'.
http://signbank.cls.ru.nl
BSD 3-Clause "New" or "Revised" License
19 stars 12 forks source link

Query, Display, Retrieval of Morphology, Morphemes #1101

Open susanodd opened 7 months ago

susanodd commented 7 months ago

What is this code supposed to do?

https://github.com/Signbank/Global-signbank/blob/da2a82ddf8bec0f489ad76f998c0dc52fec89fe8/signbank/dictionary/adminviews.py#L2323-L2325

This column of the Morphemes CSV is headed Morphology

But despite filling in all possible information for glosses and morphemes in the test dataset, this column is always empty. Also using different combinations of the code.

What is this supposed to contain?

(Perhaps when morphemes were deleted in the past, some information was never restored. See #127.)

UPDATE: @ocrasborn suspects this was part of a copy paste from a column used for CSV export of glosses. He will look at the morphemes after the holiday season. I'm working on the display of morphology so it's possible to actually retrieve what's in the database. Some examples are below.

susanodd commented 7 months ago

@ocrasborn I am condensing various templates and I came across the code for "compounds" that for some strange reason wasn't displaying anything in the Relations View. So for glosses that have or are part of compounds (based on what the code seemed to be doing), here is the repaired Relations View for the gloss Book. Does the nomenclature look okay to you?

relations-view-ngt-book-with-compounds

susanodd commented 7 months ago

I revised the Blends so that they are visible in the Relations View. There are hardly any blends in the database at the moment and it's not possible to query them. (I did this by adding some code to look for them. But that code isn't used anywhere. We probably need to add something to the query form... Need to figure out how to map this to form query fields....)

There is a blend for WEEKEND-A. So the glosses in its blend now show up here in the relations view of gloss ZATERDAG. Does the nomenclature look okay?

relations-view-blends-zaterdag

susanodd commented 6 months ago

The above screenshots are live on signbank-dev now.

susanodd commented 6 months ago

Blends query functionality added to Search. Here shown in Query View

is_a_blend_query_view

susanodd commented 6 months ago

Here is the other half, Part of a Blend

part-of-blend-query-view

susanodd commented 6 months ago

In the (very few) glosses in NGT that are blends, there is also a "role" for the blend in the morphology (not shown in the screenshots).

@ocrasborn can you study the examples and provide feedback on nomenclature / display?

susanodd commented 6 months ago

The morphology code described above has been deployed to signbank-dev.

It's possible to view the terminology there and offer suggestions for improvement here.

Because the morphology has relations between glosses going both directions, the display might be confusing.

susanodd commented 6 months ago

I found more examples of blends in other datasets and modified the display in the relations view to include the role field.

kata-kolak-relations-view-ten-B-blends

csl-shanghai-january-relations-view-blends

susanodd commented 6 months ago

Here's some more blends. (That's it.)

grammar-blends-is-wfd1975

susanodd commented 5 months ago

This is live on Signbank. Is the query mechanism sufficient?

It seems a bit weird for sequential morphology, since the "role" searched on is Component 1, Component 2, ... (It has always been this way.) I can add a Boolean "Has Sequential Morphology" instead.

The CSV routines have been revised to use the Gloss Annotation instead of the Gloss ID. This makes it more user friendly. (There is no import command for morphemes.)

ocrasborn commented 3 months ago

@susanodd Yes, the nomenclature (the use of 'compound' and 'blend' is appropriate.