Hello fellow Neo4j'ers,
for some time now I have been wondering whether it wouldn't be a nice
feature if relationship types could be arranged in a taxonomic structure,
i.e. like a class hierarchy. For example, in one of my use cases, I load
terminological terms into Neo4j. The point is, that multiple terminologies
may use the same term. Consider the following graph
Terminology1
|__ Term1
|__ Term2
|__ Term3
|__ Term4
Terminology2
|__ Term1
|__ Term3
|__ Term5
What I mean to show is that in the example, term1 has a sub term (lets say
in a IS-A relation) in terminology1 that it doesn't have in terminology2,
namely term2.
To show the different terminologies in their correct structure, e.g. on a
web site, I can do different things to sort out the issue that term1 has
different children in the two terminologies. My currently preferred
approach is to draw multiple edges between terms and their sub terms in the
following way:
Term1
|_IS-A_Term2
|_IS-A-1_Term2
|_IS-A_Term3
|_IS-A-1_Term3
|_IS-A-2_Term3
This way I know that in general, term1 has the sub terms term2 and term3.
Additionally I can tell - if I want to - that in terminology1, term2 is a
sub term of term1 (indicated by the IS-A-1 relationship type) whereas in
terminology2 this is not the case (expressed by the lack of a relationship
term1-IS-A-2-term2).
In my application, I will need both information, that is, I will want to
traverse general IS-A relationships and sometimes I will want to traverse
relationships specific to a terminology.
I immediately thought of the following:
1) build the structure
Term1
|_IS-A-1_Term2
|_IS-A-1_Term3
|_IS-A-2_Term3
i.e. only use the terminology-specific relationship types.
2) Declare:
* relationship type IS-A-1 is a subtype of relationship type IS-A
* relationship type IS-A-2 is a subtype of relationship type IS-A
Now a cypher query like
MATCH (t:TERM)-[:IS-A]-t2 WHERE t.name='term1' RETURN t2
would return term2 and term3 whereas
MATCH (t:TERM)-[:IS-A-2]-t2 WHERE t.name='term1' RETURN t2
would only return term3.
I don't know how often people could have a use for this. Currently I solve
this by using regular expressions on the relationship name like this:
MATCH (t:TERM)-[r]-t2 WHERE t.name='term1' AND r =~ 'IS-A.*' RETURN t2
but this construction depends on relationship naming and thus is neither
adequate for general use nor very robust.
Perhaps this could be a valuable extension. Do you have any thoughts
concerning this topic? Or can you actually already do this I just haven't
seen it?
Best,
Erik
--
You received this message because you are subscribed to the Google Groups
"Neo4j" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.
