Physics paper recommender

This project is part of tutorial on Link prediction with node2vec.

For this to work, you will need:

1. The MAGE graph library
2. Memgraph Lab - an application for querying Memgraph and visualizing graphs
3. gqlalchemy - a Python driver and object graph mapper (OGM)

Dataset parser

In order to parse Collaboration dataset, use public/dataset_parser.py. It assumes existance of file CA-HepPh.txt in root.

In order to run it, use following command:

python3 public/dataset_parser.py

This will prepare cypher queries which will be used in public/main.py

Link prediction script

Script in public/main.py will do the following:

• Drop database
• Import dataset from file query.cypherl prepared with public/dataset_parser.py.
• Split edges from Memgraph into test and train set
• Remove test set edges from Memgraph
• Run node2vec to get node embeddings
• Make link predictions
• Append fresh precision@k in results.txt

python3 public/main.py

Plotting results

In order to plot results use public/main.py

python3 public/plot.py