Notes.txt

These are some personal notes I'm taking while trying to understand the code.

First of all, the code is designed to work on 2 different datasets: 
- i2b2-2010 (which we don't care about) and
- SemEval 2013 (which is ours)

The file *create_partion.py* seems to be only related to the first dataset.

 The same applies to the *preprocess.py*. 
 However, this is probably where the ddi-train.pickle file should be created.
 In fact, this is what seems to be done with the other dataset.
 
 In the *svm_ddi.py* they actually train a svm on the problem! Here, they 
 actually perform the preprocessing described in the paper and this kind of 
 things. Not sure why they train the SVM tho.
 
 The file *svm-ib2b2.py* is obviously the counterpart for the other dataset.
 
 *main.py*
 - The first two lines, which seem to be never used (only in commented code)
rose an error, so I created the ./results folder. Now it works. Not sure why

- What's the ./ddi/ddi-train.pickle file?
pickle (https://docs.python.org/2/library/pickle.html , 
https://youtu.be/Pl4Hp8qwwes) serialized the object
(looks like it is frequently done with datasets for performances).
So I suppose the file is serialized in some custom format (it then reads in that 
format in the main).
This file might probably have been created in some custom file as the 
*preprocess.py*, where something similar seems to have been done for the other
dataset. The problem is: what is th exact format???
Probably we should study the *preprocess.py* and reprobuce the same format?