Best way to differentiate edges in a graph database - graph-databases

Question

I'm designing the schema for a graph database and couldn't find anything about pros and cons of how to name edges. My edges will have a minimum of two descriptors (label and property, or two properties) that describe the correct tree for a given root vertex.
here's my test graph:
// probably not efficient, but is an attempt at preventing duplicate vertices before introducing an index
g.V().has('person', 'name', 'annie').fold().coalesce(unfold(), addV('person').property('name', 'annie'))
g.V().has('person', 'name', 'bernie').fold().coalesce(unfold(), addV('person').property('name', 'bernie'))
g.V().has('person', 'name', 'charlie').fold().coalesce(unfold(), addV('person').property('name', 'charlie'))
g.V().has('person', 'name', 'danielle').fold().coalesce(unfold(), addV('person').property('name', 'danielle'))
g.V().has('person', 'name', 'edward').fold().coalesce(unfold(), addV('person').property('name', 'edward'))
g.V().has('person', 'name', 'frank').fold().coalesce(unfold(), addV('person').property('name', 'frank'))
g.V().has('person', 'name', 'gisele').fold().coalesce(unfold(), addV('person').property('name', 'gisele'))
// 1983
annie = g.V().has('person', 'name', 'annie').next()
bernie = g.V().has('person', 'name', 'bernie').next()
g.addE('manages').property('org', 'finance').property('during', 1983).from(annie).to(bernie)
// 1984
charlie = g.V().has('person', 'name', 'charlie').next()
g.addE('manages').property('org', 'finance').property('during', 1984).from(annie).to(bernie)
g.addE('manages').property('org', 'finance').property('during', 1984).from(annie).to(charlie)
// 1985
danielle = g.V().has('person', 'name', 'danielle').next()
edward = g.V().has('person', 'name', 'edward').next()
g.addE('manages').property('org', 'finance').property('during', 1985).from(annie).to(bernie)
g.addE('manages').property('org', 'finance').property('during', 1985).from(annie).to(edward)
g.addE('manages').property('org', 'finance').property('during', 1985).from(annie).to(charlie)
g.addE('manages').property('org', 'finance').property('during', 1985).from(bernie).to(danielle)
// new org
frank = g.V().has('person', 'name', 'frank').next()
gisele = g.V().has('person', 'name', 'gisele').next()
g.addE('manages').property('org', 'engineering').property('during', 1985).from(frank).to(gisele)
g.addE('manages').property('org', 'engineering').property('during', 1985).from(frank).to(danielle)
g.addE('manages').property('org', 'engineering').property('during', 1985).from(frank).to(edward)
// frank is trying his hand at finance
g.addE('manages').property('org', 'finance').property('during', 1985).from(edward).to(frank)
In regards to the edge schema in this example, the label name (I don't think...) is very important. It's already inferred who manages who by the direction of the edge. I'm not sure if there will be more than one type of edge in this graph yet, just the hierarchy. So, would there be any advantage / disadvantage to just labeling the edges with the org name or "during" year? any performance differences?
The only advantage I see so far is visual differentiation in a tool like graphexp, where it only shows the edge label...
I'm running queries like this to get the correct tree:
g.V().has('person', 'name', 'annie').emit().
repeat(outE().and(has('during', 1985), has('org', 'finance')).inV()).
path().local(unfold().values('name').fold())

Answer 1

Label is analogy to TableName in RDBMS. There are two disadvantage to "just labeling the edges with the org name or 'during' year" I can address:
In your described case, there is only one edge type. If you wanna to add another edge type like 'friendship', it will be hard to diff these two edge types. Find 'manages' edges of one person may need to iterate all edge including 'friendship' and 'manages'.
Most graph database implementations, such as 'janusgraph', hold a mapping table from 'label' string to internal label id. If there are too many edge labels, it will introduce more execution cost, or may run out of internal label id.