So Luca,  if i would for example, connect products on a store node, should 
i link 100 000 product nodes direct to a store node, creating super nodes 
and less edges for example:

(product node) SHIRT -> (store node) COOL_STORE

or fragment the model and create, character nodes  between the products and 
stores with the 2 first letters to minimize the super node, creating more 
edges like this:

(product node) SHIRT -> (character node) *S *-> (character node)  *H *-> (store 
node) COOL_STORE



What would be the better approach and why?




Em domingo, 23 de dezembro de 2012 00:00:34 UTC-2, Lvc@ escreveu:
>
> Hi Steve,
> it depends by how you access to your database. If you want to know all the 
> pages visited by a User-Agent it could make sense creating a vertex for it. 
> OrientDB is optimized for super-nodes, but sometimes the best approach is 
> to have less edges.
>
> Lvc@
>
>
> On 13 December 2012 19:39, Steve Maraspin <[email protected] 
> <javascript:>> wrote:
>
>> Hello everyone, I'm modeling a problem where my nodes are pages and my 
>> edges are visits, or better "browsing actions" from one page to another.
>>
>> I'll also need to take care of the user agent, and I was thinking about 
>> creating a dedicated property on such browsing actions (IE an edge 
>> property). I'll then need such data to filter(/aggregate?) browsing actions 
>> (IE stats). Taking this into account, I was wondering if it might be 
>> feasible/better to represent user agents as a distinct entity, rather than 
>> a property on the edge, and then have a relation between my browsing 
>> actions and user agents. Would that be good? Or would it make it a more 
>> complicated / awkward data model with no gains?
>>
>>  I think my question originates from a "normalized" way of thinking with 
>> the relational model... The point is that such "browsing actions" are edges 
>> already, so that things do look weird to me, because that'd mean that an 
>> "Edge" would connect 3 items, instead of 2 ( = not a graph anymore?).  I'm 
>> a beginner when it comes to graph data modeling, so I was curious to 
>> receive your opinion on this (also because I might have missed some other 
>> points).
>>
>>
>> Thanks a lot,
>> Steve
>>
>> -- 
>>  
>>  
>>  
>>
>
>

-- 

--- 
You received this message because you are subscribed to the Google Groups 
"OrientDB" 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/d/optout.

Reply via email to