requirements to run Magic App

[oktie]

Thanks for making your work publicly available! I just tried running the app and after finding out all the requirements one by one and installing them, I get to:

RuntimeError: Cannot open HDT file '/Users/bramsteenwinckel/Downloads/dbpedia2016-10.hdt': Not Found!

What I needed to install:

pip install streamlit
pip install stqdm
pip install semantic-ink
pip install rdflib
pip install rdflib-hdt
pip install pystardog

Perhaps you can add a requirements.txt with all the requirements?

Side note, would it be possible to pick an existing license? An open one like Apache or MIT would be great, but you can go for a ones with more restrictions, e.g. see: https://opensource.stackexchange.com/questions/11283/choosing-the-most-restrictive-open-source-license

[bsteenwi]

Hi Oktie,

Thanks for your interest in Magic.
I've updated the repository and added a requirements.txt file.

It is better to install the INK package directly from the github source using:
pip install git+git://github.com/IBCNServices/INK to get the last working version
(we are trying to fix our PyPI package asap).

The HDT DBpedia files are generated by the HDT community, this specific DBpedia version is also made available on Kaggle

The readme is updated to clarify this installation procedure.

[bsteenwi]

For the license: this is something I can't decide by myself...
I try to get in contact with our legal department.

[oktie]

Thanks for the update.

I was able to run the app now, and load a dataset. But with "start annotation" I get the following error:

TypeError: create_dataset() takes from 1 to 5 positional arguments but 6 were given
Traceback:

File "/usr/local/anaconda3/lib/python3.8/site-packages/streamlit/script_runner.py", line 354, in _run_script
    exec(code, module.__dict__)
File "/Magic/StreamlitApp.py", line 249, in <module>
    main()
File "StreamlitApp.py", line 170, in main
    annotator.annotate()
File "StreamlitMAGIC.py", line 80, in annotate
    full_emb = self.generate_embedding(tuple(all_entities),2,3)
File "StreamlitMAGIC.py", line 52, in generate_embedding
    X_train, _ = extractor.create_dataset(depth, set(data), set(), self.skiplist, jobs)

[bsteenwi]

Hi Oktie,

I think you are still using an older version of INK within the magic application.
As mentioned above:

It is better to install the INK package directly from the Github source using:
pip install git+git://github.com/IBCNServices/INK to get the last working version
(we are trying to fix our PyPI package asap).

You might need to uninstall the current INK package first (pip uninstall semantic-ink).

If you encounter additional problems, let me know

[oktie]

Actually I was going to say that I tried both and failed. But trying it again, seems like I was then missing bidict and the app wasn't starting. So I think if you add bidict to requirements.txt, this issue can be closed :)

I hope you can also replace or augment the license file with a common license.

Thanks so much!