GUI

[mshakirDr]

@elenlefoll I have added a gui template (check gui.py). You can run it in your terminal to check if it is ok. I think this should be sufficient. I have not integrated the main script to it.
Please install customtkinter to run the gui.

[mshakirDr]

GUI is functional now. Just run the gui.py. Since your directory structure is different, you can uncomment 44-64 (copied from MFTE.py) and comment 66-87. An exe can be created very easily now. Module making is another story. Probably you have to do the Mac part for exe creation as I can only do it for Windows (generally it is system dependent).

[mshakirDr]

I am gonna share a screenshot of this on my twitter. Will tag you. I hope it is ok.

[elenlefoll]

It works!!!! I did not have to install any new library, I just ran the gui.py file, it took a little while and then the white window popped up, I selected a corpus directory and it worked!!!

[elenlefoll]

The only thing is that it saves the files in rather strange places. This is probably due to differences between Windows and Mac and is why I amended all of the paths in my version of the last part of the MFTE script.
What the GUI currently does on Mac is this:
In the same directory as the one in which the directory of the corpus is found it creates a folder entitled:
CorpusDirectoryName_MFTE_Tagged_test
which itself contains

• the Stanford POS text files
• A folder entitled MD
  The folder MD then contains the MFTE tagged files and a folder called Statistics which has the three tables of count. Here is what the folder structure looks like:
  
  But anyway this is only a very minor thing that we can sort later. You certainly deserve your break!

[elenlefoll]

Since your directory structure is different, you can uncomment 44-64 (copied from MFTE.py) and comment 66-87.

I didn't read this bit properly! Sorry, I'll try again! I was too excited to try it 😄

[elenlefoll]

Success! 🥳

[mshakirDr]

I have refined the button code a bit. Now it will run no matter how many times button is clicked, and no execution if file dialogue is escaped or closed without a directory.
The directory structure part is now in call_MFTE (yours above). At some point this should be 1. I don't like 3 separate folders. MAT style directory structure looks good to me: 1 folder then another inside it and the last one Statistics with all output files.

[elenlefoll]

I'm not a fan of the MAT structure but perhaps we can strike a compromise? My suggestion would be:

• Creating a single folder in the same directory as the corpus: CorpusDirectoryName_MFTE
• which itself contains three folders:

1. StanfordPOS_Tagged > with the files as tagged by the Stanford Tagger
2. MFTE_Tagged > with the files as tagged by the MFTE
3. Statistics > with the three tables of frequencies

What do you think?

[mshakirDr]

I have updated the code.

[mshakirDr]

I have spent 5 or so hours today to create an exe. I can build a GUI plus console-based app which loads in two windows and works fine. However, still I am not the able to create a standalone GUI executable file. Somehow the tagger stanza is messing with the GUI. I have asked for help on Stackexchange, let's see if we find a solution.

[mshakirDr]

@elenlefoll A working .exe is now possible for Windows. Mac OS version will be your responsibility.

[elenlefoll]

That's absolutely brilliant, Shakir! Well done and thank you!

Unfortunately, I do not currently have the skills to create an executable for MacOS nor the time to develop them, so this will have to be a future project, perhaps for potential future users of the tagger, if they are interested in this. For now, my next priority, in parallel to continuing the slow and tedious evaluation work, is to draft a paper for submission to a journal so that we have a citable tagger and can hopefully "capitalise" on our hard work!

[mshakirDr]

Unfortunately, I do not currently have the skills to create an executable for MacOS

It is not that difficult. Basically you will just clone the current python environment using a GUI python module. I will make a small guide for you once the release is near.

[mshakirDr]

Tagger can be cited as a software too, so I say we release it as a beta and let people use it, until a paper is submitted and published. It is going to take six months to a year for such a paper to publish.

[elenlefoll]

Tagger can be cited as a software too, so I say we release it as a beta and let people use it, until a paper is submitted and published. It is going to take six months to a year for such a paper to publish.

Sure, I agree, but I'd still like to work on the paper asap because we can also publish that as a pre-print and thus attempt to accelerate the process somewhat because, you're right, academic publishing is sooooooooo frustratingly slow.