Retrofitting
This code is used for improving document embedding obtained from PV-DBOW by adding citation information. Our 
experimentation shows that the addition of network information improves classification score by atleast 5% to 7%.

Requirement: Python 2.7

Data: 
idxs.txt: contains ids of the documents
id1
id2

labels.txt: contains labels of the documents
words.txt: contains content of the document

Each line of idxs,labels and words file contain information of the same document

#Running the program
Run the following command:
python Run.py

This will produce d2v embs, it will then retrofit it. Along the way it will also do classification and visualize the embeddings using PCA.