Predict Future Connections between Facebook Pages


Original dataset: http://networkrepository.com/fb-pages-food.php


Use neo4j database 3.5.17.
Install APOC and Graph Algorithms plugin.

1) Create user:

LOAD CSV WITH HEADERS FROM "file:///fb-pages-food.csv" AS line
CREATE (:user { id: line.id, name: line.name, new_id: line.new_id})


2) Create: FOLLOWS

LOAD CSV WITH HEADERS FROM "file:///connect.csv" AS row
MERGE (u1:user {new_id: row.from})
MERGE (u2:user {new_id: row.to})
MERGE (u2)-[r:FOLLOWS]->(u1)

3) Create TRAIN:
LOAD CSV WITH HEADERS FROM "file:///test.csv" AS row
MERGE (u1:user {new_id: row.from})
MERGE (u2:user {new_id: row.to})
CREATE (u2)-[r:TRAIN {flag: row.flag}]->(u1)

4) Create TRAINVEC:

LOAD CSV WITH HEADERS FROM "file:///train2.csv" AS row
MERGE (u1:user {new_id: row.from})
MERGE (u2:user {new_id: row.to})
CREATE (u2)-[r:TRAINVEC {flag: row.flag}]->(u1)

Run fb.ipynb to evaluate the result.