[Nanocubes-discuss] Building the index: example with more than one categorical dimension

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

[Nanocubes-discuss] Building the index: example with more than one categorical dimension

enetsee
Hi, 

I recently came across the nanocubes paper and would like to understand better the process for building the index. 

The paper contains a worked example with one spatial dimension, one categorical dimension and a temporal dimension. 

From what I can understand, the indexing process involves building nested tree data structures for the spatial and categorical dimensions with the final temporal dimension using the summed-area table variant, with memory efficiencies gained by sharing subtrees (and template based container specialization, based on the cardinality of each dimension). 

Would I be correct in thinking that, for two categorical dimensions, the flat-tree representing the first categorical dimension would simply point to further flat-trees representing the second categorical dimension? 

Is there an example anyone could share, similar to that given in section 4.2 of the paper, which shows how the index is built in this case? 

Many thanks, 

Michael 

_______________________________________________
Nanocubes-discuss mailing list
[hidden email]
http://mailman.nanocubes.net/mailman/listinfo/nanocubes-discuss_mailman.nanocubes.net
Reply | Threaded
Open this post in threaded view
|

Re: [Nanocubes-discuss] Building the index: example with more than one categorical dimension

laurolins
Hi Michael,


The paper contains a worked example with one spatial dimension, one categorical dimension and a temporal dimension. 

From what I can understand, the indexing process involves building nested tree data structures for the spatial and categorical dimensions with the final temporal dimension using the summed-area table variant, with memory efficiencies gained by sharing subtrees (and template based container specialization, based on the cardinality of each dimension). 

You are right. Sharing whole trees (from the next dimension) or branches (from the same dimension) is the key for saving memory while encoding a data cube.

Would I be correct in thinking that, for two categorical dimensions, the flat-tree representing the first categorical dimension would simply point to further flat-trees representing the second categorical dimension? 

Exactly.

Is there an example anyone could share, similar to that given in section 4.2 of the paper, which shows how the index is built in this case? 


Here are images for the same example from the paper with an extra categorical dimension (one extra flat-tree layer).


Thanks,
Lauro



On Feb 3, 2015, at 6:44 PM, Michael Thomas <[hidden email]> wrote:

Hi, 

I recently came across the nanocubes paper and would like to understand better the process for building the index. 

The paper contains a worked example with one spatial dimension, one categorical dimension and a temporal dimension. 

From what I can understand, the indexing process involves building nested tree data structures for the spatial and categorical dimensions with the final temporal dimension using the summed-area table variant, with memory efficiencies gained by sharing subtrees (and template based container specialization, based on the cardinality of each dimension). 

Would I be correct in thinking that, for two categorical dimensions, the flat-tree representing the first categorical dimension would simply point to further flat-trees representing the second categorical dimension? 

Is there an example anyone could share, similar to that given in section 4.2 of the paper, which shows how the index is built in this case? 

Many thanks, 

Michael 
_______________________________________________
Nanocubes-discuss mailing list
[hidden email]
http://mailman.nanocubes.net/mailman/listinfo/nanocubes-discuss_mailman.nanocubes.net


_______________________________________________
Nanocubes-discuss mailing list
[hidden email]
http://mailman.nanocubes.net/mailman/listinfo/nanocubes-discuss_mailman.nanocubes.net