Cypher Queries for Recommendation System for Predicting Links between Companies

I have my dataset on Neo4J and am looking to develop a recommendation system for predicting links between companies. I did go through the data science course on Neo4J online training academy and I am trying to replicate the same for my dataset (https://colab.research.google.com/github/neo4j-contrib/training-v2/blob/master/Courses/DataScience/notebooks/04_Predictions.ipynb#scrollTo=LXPKhjebvR7o). However, I am not sure what cypher queries will be suitable. I did try to build the system by splitting the data on the basis of date of incorporation.

So far I have used the following cypher queries:
MATCH (c:Company)
WHERE c.incorporation <> "NA"
RETURN right(c.incorporation, 4) as year, count(c) as numCompanies
ORDER BY year DESC

MATCH (c:Company)
WITH right(c.incorporation, 4) as Year
WHERE Year <"2011"
MERGE (a)-[:INC_EARLY {year:Year}]-(d)

MATCH (c:Company)
WITH right(c.incorporation, 4) as Year
WHERE Year >"2011"
MERGE (a:Company)-[:INC_LATE {year:Year}]-(b:Company)

MATCH ()-[:INC_EARLY]->()
RETURN count(*) AS count

MATCH ()-[:INC_LATE]->()
RETURN count(*) AS count

I am not sure which queries or which information can be used to build the recommendation system. I thought of using the concept of time to have a better data split.

Any guidance would be of great help.