That helps.  Thanks!

Scott

On Friday, May 8, 2015 at 3:47:03 PM UTC-4, Scott Orr wrote:
>
> My team is setting up a database for an analytical layer of a website. 
>  Our goal is to store large amounts of data related to our users and 
> inventory, such as information on browsing sessions, user feedback, tweets 
> and Facebook posts, etc.  We want to be able to use that information to 
> make recommendations to users in real time (e.g., Amazon-type stuff, like 
> "Other people who viewed this item also viewed....").
>
> We're using Orient 2.0.8.
>
> From looking at the Orient wiki and the questions answered on this and 
> other groups, we think we need to be using LINKLISTs for part of our 
> application.  For example, we'd like to be able to store tweets about each 
> of the 20,000 items in our inventory.  Ideally, the record for each item 
> would include links to each of those tweets, which could number in the tens 
> of thousands--and a LINKLIST seems like the obvious way to bundle all of 
> those links into a single property, with the added bonus that a LINKLIST is 
> ordered, which means that, among other things, we could pop the last 10 or 
> so tweets about a given item, and display those on the page describing that 
> item's details.
>
> My first question is, can a LINKLIST be used with bidirectional edges (I 
> assume they'd be lightweight edges)?  We need to be able to traverse the 
> links in both directions.
>
> If LINKLISTs must consist only of unidirectional links, we'll have to 
> create links at end end of each pair (so, a LINKLIST pointing out of an 
> inventory item, to all of its related tweets, and a link from each tweet 
> back to that item).  If we have to do this, can we still use the Graph API 
> to traverse those links, and if not, is link traversal any slower with the 
> Document API?
>
>
> Thanks.
>

-- 

--- 
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