I have noticed that there are no pre-trained Maltese ConceptNet Numberbatch embeddings available in the repository yet, and I need them for my university project. I would like to try to train them myself. I have a few clarification questions on how the Numberbatch embeddings in other languages were created. I read the ConceptNet paper and as far as I understand, the average over Word2Vec, Glove and ConceptNet embeddings represents the Numberbatch. Correct me if I am wrong, please. Also, for training the ConceptNet embeddings, you just used the edges as the context, right? Is there any code which I can use for training the ConceptNet embeddings.
Thank you in advance!