pretrained_word_embeddings.py:
This file will use pre-trained word embeddings to prepare representations of the argument and the wikipedia articles, then do a sum over the dot product of the representation of the argument with each wikipedia article and finally use the sum as the representation of argument combined with knowledge from wikipedia. This final combined representations of two articles will be used to do classification between the two for which one is more convincing using logistic regression.