Neo4j - data modeling - neo4j

Question

I am building some test application with Neo4j. I want to model small social network and try to find:
All friends of user X
Friends of user's X friends, who likes beer
I stuck with modeling "know" relation. Let's take 3 users A, B and C. Is it enough to define only one relation between, them ex.
A knows B
B knows C
or I have to make 'bidirectional' relations and explicitly provide that
A knows B so B knows A
B knows C so C knows B
it will increase the number of relations, but maybe it is needed.
The same question is related to favorite drink.
A likes beer
should I also define?
beer is liked by A

Answer 1

If you want to be able to differentiate between a knowing b and b knowing a, then you need to have two relationships. Otherwise, at query time, you can easily get both by omitting the direction.
Similarly, with beer being liked, you really only need to define one direction.
For a real example: The facebook/linkedin model where connections are mutual only needs one direction/relationship, but the twitter model where one person can follow another (but the other person doesn't have to follow them back), you'd need two relationships--one for each direction.
Update with some query examples:
CREATE
(joe {name:"Joe"}),
(jim {name:"Jim"}),
(bob {name:"Bob"}),
(beer {name:"Beer"}),
joe-[:friends_with]-jim,
joe-[:friends_with]-bob,
bob-[:likes]->beer;
For the friends:
START person=node:node_auto_index(name="Joe")
MATCH (person)-[:friends_with]-(friend)
RETURN person, friend;
And the friends who like beer:
START person=node:node_auto_index(name="Joe"), beer=node:node_auto_index(name="Beer")
MATCH (person)-[:friends_with]-(friend)-[:likes]->(beer)
RETURN person, friend, beer;