Literature model

[danich1]

This PR implements and compares the CoCoScore text mining system against my current system. I used hetionet v1 edges for the comparison, where the positives are edges already established in hetionet and the negatives are edges not in hetionet v1. There might be a might be a confounding issue for precision-recall performance, since hetionet v1 is not a fully complete network (some non-zero edges might actually be true edges but are considered negative in this context).

[danich1]

So for the .png file, you're assuming CoCoScore is the baseline that you want to compare you're relationships to?

The .png file compares our model with the CoCoScore literature model. For this particular instance CoCoScore is our "baseline" model. The goal here is to show that our method performs just as good as a literature model.