You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@giraph.apache.org by Ahmet Emre Aladağ <em...@agmlab.com> on 2013/07/17 08:38:25 UTC
Pipes of computation
Hi,
Is there a way to pipe the Computation classes?
Example: Run PageRankComputation and then run NormalizationComputation
on the same graph without activating Output/Input formats between the two.
This way we can re-use some computations like log-scale CDF
normalization and don't mess up the original PageRank computation.
Re: Pipes of computation
Posted by Ahmet Emre Aladağ <em...@agmlab.com>.
Thank you, it seems that can work. But it requires us to count the
supersteps of two computations combined. It could have been nice if that
was loosely coupled such that in the second computation I could check if
that's the 0th superstep or not.
On 07/17/2013 11:08 AM, Han JU wrote:
> I think that can be achieved using MasterComputation.
> Take a look at o.a.g.master.TestSwitchClasses
>
> Hope that helps.
>
>
> 2013/7/17 Ahmet Emre Aladağ <emre.aladag@agmlab.com
> <ma...@agmlab.com>>
>
> Hi,
>
> Is there a way to pipe the Computation classes?
>
> Example: Run PageRankComputation and then run
> NormalizationComputation on the same graph without activating
> Output/Input formats between the two.
>
> This way we can re-use some computations like log-scale CDF
> normalization and don't mess up the original PageRank computation.
>
>
>
>
>
> --
> *JU Han*
>
> Software Engineer Intern @ KXEN Inc.
> UTC - Université de Technologie de Compiègne
> ////GI06 - Fouille de Données et Décisionnel//
>
> +33 0619608888
Re: Pipes of computation
Posted by Han JU <ju...@gmail.com>.
I think that can be achieved using MasterComputation.
Take a look at o.a.g.master.TestSwitchClasses
Hope that helps.
2013/7/17 Ahmet Emre Aladağ <em...@agmlab.com>
> Hi,
>
> Is there a way to pipe the Computation classes?
>
> Example: Run PageRankComputation and then run NormalizationComputation on
> the same graph without activating Output/Input formats between the two.
>
> This way we can re-use some computations like log-scale CDF normalization
> and don't mess up the original PageRank computation.
>
>
>
--
*JU Han*
Software Engineer Intern @ KXEN Inc.
UTC - Université de Technologie de Compiègne
* **GI06 - Fouille de Données et Décisionnel*
+33 0619608888