Make more information of gloss publicly accessible

[rem0g]

For example https://signbank.cls.ru.nl/dictionary/gloss/729 only show video and senses, it's too short of information for non logged in users.

The page should display relation with other glosses, such as variants (METRO-A and METRO-B).
Phonology
Voorbeeldzinnen
NMM videos
And finally 3-perspective video

@uklomp let me know if you agree with this.

[uklomp]

There is indeed a difference between the public version and the gloss-detail page (last one only visible for logged in users). As I understand it, you don't want this difference anymore Gomer? I'd like to know why this difference was there in the first place, maybe there was a good reason for this? @ocrasborn

[uklomp]

I will discuss this together with Floris and Marloes - on the one hand reconsidering what should be publicly available, on the other hand also keeping in mind that colleagues might be developing a more public friendly interface anyway.

[susanodd]

It's fairly easy to add the requested information.

Since the non-NGT datasets have their own permissions, and since each gloss can be visible or not (In Web setting), then the information that's shown can be controlled that way.

The new Synset Wordnet information might be nice in the public view. There are also social media links in the public view.

[rem0g]

I will discuss this together with Floris and Marloes - on the one hand reconsidering what should be publicly available, on the other hand also keeping in mind that colleagues might be developing a more public friendly interface anyway.

Not to start a discussion here; but since Signbank is a scientifically lexicographically dictionary subsided by the dutch government and they are all available on the Signbank commercially-free; it would not make sense not to publish all relevant details if you get my gist?

I also want to know opinion from @ocrasborn

[ocrasborn]

Hi all, what is in the public view now was not carefully chosen by me, we simply copied the Auslan Signbank public view. I agree with @rem0g that it would be appropriate to make much more ‘public’ in the sense of easily accessible for anyone, in particular the language community.

What kept me back in the past is simply that it’s not easy to develop an interface where everything speaks for itself, we simply didn’t invest in that until now. See however the pilot version of an app developed by @TobiasCunnen , which is much nicer for lay users than simply a list of text values

[uklomp]

I will discuss this together with Floris and Marloes - on the one hand reconsidering what should be publicly available, on the other hand also keeping in mind that colleagues might be developing a more public friendly interface anyway.

Not to start a discussion here; but since Signbank is a scientifically lexicographically dictionary subsided by the dutch government and they are all available on the Signbank commercially-free; it would not make sense not to publish all relevant details if you get my gist?

I also want to know opinion from @ocrasborn

I completely agree about the subsidy etc.. but at the same time I'm thinking: having an account is not really difficult right? You only need an email address. People get access automatically so I dare say it's quite publicly available. And some might say that the simple overview is actually more user-friendly.

BUT I also see that we agree that at least for NGT, most (all?) details from the 'details' page could be made publicly available. The question is now if we should consult the other dataset managers first, to implement this for the whole website, and if we want the 'public view' and 'details' to be more alike, or if one of the two pages is superfluous now... also, since we will build another platform together with TYD, should we focus on that instead? These are the questions I want to discuss with Marloes and Floris but feel free to write your opinion here!

[susanodd]

The primary difference is the EDIT capability on the Gloss Detail. There is a huge amount of code related to this. The Gloss Edit page has 1000 lines of code just for the Senses, to be able to edit them.

It's straightforward to add the information requested by @rem0g to the public Gloss template page.

CODE: the template page for public view needs to remain separate from the template with Edit in the implementation. This has to do with all the urls to update the gloss in the database. Permissions are needed for each of these.

For the public template, Django does not need to generate any code in the browser for the "edit" related code.

The Gloss Edit template has 3000 lines of code. Plus numerous javascript files.

APPOLOGIES FOR THIS PARAGRAPH, it has to do with "Django processing the template to generate the html": To reduce the amount of code, it has not been possible to put the code into separate templates to load. The translations inside of the loaded templates weren't processed. (There have been various issues for this in the past, but nobody solved this.) This has been successful with AJAX calls that insert the results into the page.

UPDATE: I implemented this as described by the issue. See screenshots below and pull request.

[susanodd]

I implemented this.

There is a pull request.

It's not on any development server yet.

Because of some details with permissions and new code for serving the new kinds of filenames for the NME, perspective, sentence videos to anonymous users, it would be most helpful to have some actual videos on one of the development servers in order to see things as an anonymous user.

To develop the code, on my own computer, I needed to make the test dataset public and some glosses there be public and add all the kinds of videos in order to browse them and see the new public view.
But this is still different than actually browsing the real NGT dataset as an anonymous user.
I fixed a number of errors for anonymous users. (See the pull request.)

[susanodd]

Screenshot

[uklomp]

Can you please wait until we made a more concrete decision, before you do any work that needs to be undone? I will get back to it this week.

[susanodd]

Can you please wait until we made a more concrete decision, before you do any work that needs to be undone? I will get back to it this week.

I implemented it now, because per January 1, my hours are being reduced to 50%. This means more work for the others. I want to make sure it gets done.

Nothing will need to be undone. It's all modular.

The main changes are to allow the "other" kinds of videos to be displayed anonymously. Nothing is required. But if an anonymous user has obtained a link to one of the "other" videos, the software won't crash now. This has nothing to do with the display.
Also, the existing search functionality is broken on signbank. It keeps changing to NGT and won't let users search other datasets.

[uklomp]

Questions to discuss on Thursday:

1. would it be possible to make changes for public / detailed view for only specific datasets?
2. would it be easiest to have the exact same lay-out for the things that would be visible on both pages?
3. how much time would it cost to implement it in the way Gomer proposed? or actually: is this feasible to ask in the upcoming months or is it simply too much?

[susanodd]

(1) yes (can add settings fields to Dataset class as is done for field choice hiding)
(2) yes, this is straightforward (I started with the existing layout, so can just adjust the html blocks)
(3) I already implemented what Gomer proposed (but layout needs to follow 2)

To achieve (2) and make it seamless looking, for a "logged in used" in Details views, it needs to do "permission checks for anonymous users" on the Public view rather than just show it. (Since anonymous users might not be able to view it.)

For (1) this is a really nice idea. Then the dataset manager has more freedom. The field choice exclusion was a bit buggy and isn't maintained. We can just fetch the appropriate setting for each of the items Gomer suggested and then hide or show for the dataset as per settings.

[susanodd]

@uklomp I revised the screenshot above.

[uklomp]

Okay, our conclusion is that if it's indeed "that easy" we would like to have the information that Gomer proposed visible in public view as well, but only for the NGT dataset and only for signs that are "in the dictionary". In the screenshot above, I noticed the following things:

• We'd like the morphology panel there as well
• The panel of the NME video's has a different name here than on the detailed gloss page; it should be the same
• The panels of "relaties met andere gebaren" and "aantekeningen" should not be there

[susanodd]

LAYOUT PUBLIC VIEW PANELS

@uklomp One layout question, for "empty" panels, do you want the panel title to show? Or for it to be hidden?

(For display consistency)

On the detail view, there is a tiny glyphicon on top of the panel heading if there is data.

(The glyphicons may not be consistent? New kinds of data didn't add this. We can also entirely hide the panel heading if there's no data. Depending on the settings for the different datasets public data, then we would hide (not generate) the data.

The panels can be "open" on page load, depending on your preference. Just say which ones.

[susanodd]

SEARCH BY MORPHOLOGY

Semi-related to public view, since the user only finds published glosses

Should the search fields be altered to make them more accessible?

It took me a long time to find:

example of Simultaneous Morphology:
morpheme onder-boven
gloss BINNEN-A

To find this in the interface under Search By Morphology, you need to "guess" correctly that the morpheme is "onder-boven" to search for it. Nearly all of the morphemes have no glosses they appear in.

If I remember correctly, @ocrasborn previously said that there should be about 400 glosses with morphemes.

I think the search needs to find "anything that has a simultaneous morphology" and not need the user to choose a morpheme.

1. There are minimal results if you search by "is a blend" or "is part of a blend"

2. For sequential morphology, there are the "component N" choices. This is kind of ineffective. This could be changed to "has sequential morphology"

3. For the morpheme type, this assumes the Morpheme objects have been used. There are virtually no glosses that use them.

(1) For a researcher, it is handy since if there are glosses the user is aware of, they will know if morphology is missing. Only two NGT glosses are blends (at the moment). For an anonymous user, it kind of looks like the feature might be broken if there are minimal results?

(2) For sequential morphology, since it is the combination of glosses, there is no phonology. Should this be displayed in a different way? In the Gloss Details View, the morphology section is set up for the purpose of editing. It looks a bit "computery" to me, with labels "component 1", "component 2". For a casual user, maybe it should be displayed differently?

(3) For the morpheme type, this is specific to Morpheme objects. Do we want the morpheme objects to be visible anonymously? (It's quite difficult to actually find any glosses with "simultaneous morphology". Only a couple of the morpheme types are actually used (@ocrasborn ?)

[susanodd]

RELATIONS

For variants, the way this is implemented, there is a difference between "pattern matched" versus "relation to other sign".

It was implemented in the past that the user may specify if the signs are related.
If a sign is related to another sign explicitly, then the pattern match isn't shown. (The explicit relations override the inferred relations.)

For example, METRO-A, METRO-B. (pattern variant)

STEKKER-A, STEKKER-B (synonym)

(I don't know whether this is still the intention. It was implemented that way as per @ocrasborn )

(I am looking up searches in the public view to make sure the things searched on show up in the view.)

[susanodd]

Okay, our conclusion is that if it's indeed "that easy" we would like to have the information that Gomer proposed visible in public view as well, but only for the NGT dataset and only for signs that are "in the dictionary". In the screenshot above, I noticed the following things:

• We'd like the morphology panel there as well
• The panel of the NME video's has a different name here than on the detailed gloss page; it should be the same
• The panels of "relaties met andere gebaren" and "aantekeningen" should not be there

The "relations with other signs" is the "METRO-A" and "METRO-B".

"Aantekeningen" will disappear.

The NME Videos translation is because I'm running it on PyCharm locally, with an outdated translation file. I'll fix that when I get it working on the development server.

other stuff

I'm hiding things from the Gloss Search for Public searching when the things can't be displayed in the Public View. (Otherwise it makes no sense to be able to search on them.) (The Gloss Search on the master server does not distinguish between anonymous or authenticated for the search form. Or rather, only publication details are not shown)

If you do some searching as anonymous, one can see if interface things don't make sense.

[uklomp]

LAYOUT PUBLIC VIEW PANELS

@uklomp One layout question, for "empty" panels, do you want the panel title to show? Or for it to be hidden?

(For display consistency)

We want this the same as for the detailed view, which is that the panel title is always visible, right?

[susanodd]

This has been updated on signbank-susan

All the publicly viewable panels are shown, even if empty.

The sentences are NOT visible now.

Scrolling around as an anonymous user works.

There are hardly any videos on the server.