This version is not official.
Conda always installed the stable version, even if I was asking for the latest!!!
Therefore, I had to manually download the latest into my local directory.
Anyway, it still gave me problems... maybe, being the latest, it had some bugs.
In any case, since the problematic code was related to things that I did not need at the time (in particular: built-in datasets), I simply DELETED it.
If you really need the code related to the built-in datasets (as FB15k-237, FB15K, wiki-something etc) you have two possibilities:
1- Download them separately and then process them as external datasets (some additional work is needed in this case)
2- Try to download the latest version and use it (maybe, at the time you are reading this message, all works well finally)
In particular, I had these problems in March 2022. You can check if the repo has been updated in the meanwhile.
For my pc the following works (You can change the name of the env in the first line of environment.yml):
conda env create -f environment.yml
conda activate NAME
git clone https://github.com/FMagnani/dgl-ke.git
cd dgl-ke/python
python3 setup.py install