Potential Performance Improvements

[kkaris]

The neo4j python documentation have a page with performance recommendations, some of which are worth a try.

The ones that look relevant for us are:

Specify the target database on all queries when creating a new session:

driver.session(database="<DB NAME>")

Status: will not implement (some queries ran faster, but most queries actually ran slower with specifying the database by a small margin)

Use query parameters instead of hardcoding or concatenating values into queries:

session.run("MATCH (p:Person {name: $name}) RETURN p", name="Alice")

Status: proposed in #129, however, the speedup was not very big and not all queries tested actually became faster although the queries with speedup were in the majority.
Status: added in #129

Always use node labels:

session.run("MATCH (p:Person|Animal {name: $name}) RETURN p", name="Alice")

Use multithreading (or async if we upgrade to 5.x) to run concurrent sub-queries on a complex query e.g. get_stmts_for_stmt_hashes and queries depending on it.

[bgyori]

These are pretty good, I think doing the first one is trivial since session creation is done in a single place. The second one should also be straightforward, we just need to modify the low-level Cypher queries to be constructed this way instead of string manipulation. As for node labels, I think we are already doing that whenever possible. The last one could also be interesting, you could try it out on get_stmts_for_stmt_hashes to see if it makes a difference.