Hi Toby,
What was your final design and did you use orientdb?

2 Eylül 2012 Pazar 21:08:01 UTC+3 tarihinde Lvc@ yazdı:
>
> Hi Toby,
> The graph approach you found is reasonable but you could use indexes to 
> speed up lookups in tags. But if you have few tags (max 10 ad you reported) 
> then probably the index doesn't speeds up things.
>
> Lvc@
>
> Sent from Samsung Galaxy 2
> Il giorno 02/set/2012 15:39, "FieldMarshallVague" <[email protected] 
> <javascript:>> ha scritto:
>
>> Hi there,
>>
>> I'm new to OrientDB, but I really like what I'm seeing and think that it 
>> looks like the best solution to a product catalogue that I'm building.  I 
>> have read so much documentation, but I think I am starting to confuse 
>> myself now.. so perhaps someone can help?
>>
>> I want to store product-level tags against a very large catalogue (1M+ 
>> items).  The number of tags against each item is unlimited, but many will 
>> share the same tags and most will have at least one tag in common.  There 
>> is no theoretical limit on the numbers of tags used to search, but I expect 
>> this to be around 5-10 on average.
>>
>> For instance, 800,000 will be tagged 'plastic', while 300,000 will be 
>> tagged 'metal'.  Obviously, if I keep the tags at product-level (as an 
>> associative array/embedded list within the record itself) then there will 
>> be high levels of duplication and searching will be very slow (this is an 
>> assumption, I haven't benchmarked).  So what is the best way to store the 
>> tags so that I can search very fast with multiple-tag criteria?
>>
>> I was thinking of using vertices for the tags and creating edges 
>> connecting to products with that tag.  But I am not sure if this is a fast 
>> method or how exactly I would attach the graph-data to the product data 
>> record that contains the name, price, description etc.   Is this the right 
>> approach?
>>
>> Some simple questions that may help define the problem:
>>
>> Would a vertex with 800,000 in (or out) edges be a problem for Orient 
>> server on a high-spec PC?  i.e.  Would a query to retrieve items tagged 
>> 'plastic' (800,000), 'energy saving' (20,000), 'white' (30,000) and 
>> 'adaptor' (1,000) be horribly slow?  
>>
>> Is it really possible to store my product data as document-style records 
>> while connecting them to vertex-tags via the graph?  It seems like the 
>> graph and document concepts are both types of records anyway, but I am just 
>> checking that I am not being really stupid.
>>
>> Sorry for being such a dumb-ass, I have read the docs exhaustively but 
>> I'd really appreciate some insight here before I continue.
>>
>> Thanks very much in advance!
>>
>> Toby.
>>
>>
>>
>> -- 
>>  
>>  
>>  
>>
>

-- 

--- 
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