You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mahout.apache.org by Andrew Wang <an...@gmail.com> on 2009/11/17 10:16:30 UTC
Re: [jira] Commented: (MAHOUT-18) Embrace interoperability with other
softwares
Sorry to disturb you guys! I am data mining programmer using WEKA(
http://www.cs.waikato.ac.nz/ml/weka/) for several years. This funcation may
be very useful for us who'd like to import the trained model in WEKA into
Mahout.
Look forward to it.
On Tue, Nov 17, 2009 at 5:02 PM, Ted Dunning (JIRA) <ji...@apache.org> wrote:
>
> [
> https://issues.apache.org/jira/browse/MAHOUT-18?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12778781#action_12778781]
>
> Ted Dunning commented on MAHOUT-18:
> -----------------------------------
>
>
> This will be important someday.
>
> At that time, we should open a new JIRA and implement it. Right now, we
> are working on getting relevant capabilities. Until we have them,
> interchange is fruitless.
>
> > Embrace interoperability with other softwares
> > ---------------------------------------------
> >
> > Key: MAHOUT-18
> > URL: https://issues.apache.org/jira/browse/MAHOUT-18
> > Project: Mahout
> > Issue Type: New JIRA Project
> > Reporter: Shunkai Fu
> > Priority: Minor
> >
> > This is an open issue. It is related with all possible components
> existing or to born in the future.
> > ML or DM models normally have two phases: training and scoring (or
> predicting). If we agree "updating" is an independent one, we will have 3
> phases.
> > There are many softwares about ML/DM outside. We want the users of Mahout
> be able to import models got built from other software here, update them
> and/or use them for scoring. To achieve this goal, we need to recognize the
> commonly used formats.
> > Besides, users may choose Mahout because Mahout is speedy in learning.
> After a model is ready, they may export the model trained, view it with some
> visualization tool, or import it into other software or application for
> scoring (or predicting). In this case, exporting into widely recognized
> format is expected.
> > Finally, I want to say that the importing and exporting will not
> influence the ongoing projects, so developers of other components need not
> worry about this.
>
> --
> This message is automatically generated by JIRA.
> -
> You can reply to this email to add a comment to the issue online.
>
>
Re: [jira] Commented: (MAHOUT-18) Embrace interoperability with other
softwares
Posted by Robin Anil <ro...@gmail.com>.
Weka has its training and classification algorithm and many of them.
Infact, currently mahout only have the Naive Bayes implementation and
a improved version of the same. What i am saying is, the model trained
using Naive Bayes or a classifier where classification reduces to
multiplying the document vector with various label vectors in a model
could be imported(atleast theoretically) from Weka to Mahout. I am not
sure that is true for an SVM model or something that Mahout doesnt
have but Weka has. But its a nice suggestion. Could take a look at
Weka data format and see.
Robin
On Tue, Nov 17, 2009 at 3:14 PM, Andrew Wang <an...@gmail.com> wrote:
> As you know, i am new guy about the Mahout. suppose i have one model trained
> in WEKA using distinct classifiers, if the Mahout have some port to import
> the model, and using the model in the up-coming process, it will be very
> cool.
> Sorry to make you confuse!
>
> On Tue, Nov 17, 2009 at 5:20 PM, Ted Dunning <te...@gmail.com> wrote:
>
>> Can you say more about what you really need?
>>
>>
>>
>> On Tue, Nov 17, 2009 at 1:16 AM, Andrew Wang <andrew.wang.1900@gmail.com
>> >wrote:
>>
>> > Sorry to disturb you guys! I am data mining programmer using WEKA(
>> > http://www.cs.waikato.ac.nz/ml/weka/) for several years. This funcation
>> > may
>> > be very useful for us who'd like to import the trained model in WEKA into
>> > Mahout.
>> > Look forward to it.
>> >
>>
>>
>>
>> --
>> Ted Dunning, CTO
>> DeepDyve
>>
>
Re: [jira] Commented: (MAHOUT-18) Embrace interoperability with other
softwares
Posted by Ted Dunning <te...@gmail.com>.
On Tue, Nov 17, 2009 at 3:28 AM, Robin Anil <ro...@gmail.com> wrote:
> On Tue, Nov 17, 2009 at 3:55 PM, Isabel Drost <is...@apache.org> wrote:
> > On Tue Andrew Wang <an...@gmail.com> wrote:
> >> As you know, i am new guy about the Mahout. suppose i have one model
> >> trained in WEKA using distinct classifiers, if the Mahout have some
> >> port to import the model, and using the model in the up-coming
> >> process, it will be very cool.
> >
> > Could you please explain exactly which models you would like to import
> > and why?
> >
> > Assuming we are talking about naive bayes: What is really expensive
> > about it is training the classifier. I wonder why you would want to do
> > that within Weka.
> I can see a use case where a small model is used to classify a large
> set of documents in parallel.
>
>
One case is to compare training algorithms.
Another is to deploy a model with known characteristics in a new production
setting.
Re: [jira] Commented: (MAHOUT-18) Embrace interoperability with other
softwares
Posted by Robin Anil <ro...@gmail.com>.
On Tue, Nov 17, 2009 at 3:55 PM, Isabel Drost <is...@apache.org> wrote:
> On Tue Andrew Wang <an...@gmail.com> wrote:
>> As you know, i am new guy about the Mahout. suppose i have one model
>> trained in WEKA using distinct classifiers, if the Mahout have some
>> port to import the model, and using the model in the up-coming
>> process, it will be very cool.
>
> Could you please explain exactly which models you would like to import
> and why?
>
> Assuming we are talking about naive bayes: What is really expensive
> about it is training the classifier. I wonder why you would want to do
> that within Weka.
I can see a use case where a small model is used to classify a large
set of documents in parallel.
>
> With most classification algorithms I am familiar with, training is
> expensive, but application to new instances is cheap. That is why
> currently I do not really understand, why you would want to run the
> training in Weka and use the model in Mahout. However, I could imagine
> use cases where you might want to train the model with Mahout and use
> it as part of a processing chain within Weka.
>
>
> Isabel
>
Re: [jira] Commented: (MAHOUT-18) Embrace interoperability with other
softwares
Posted by Andrew Wang <an...@gmail.com>.
Isabel, your comments make sense. Maybe my thought is childish, anyway,
thks!
On Tue, Nov 17, 2009 at 6:25 PM, Isabel Drost <is...@apache.org> wrote:
> On Tue Andrew Wang <an...@gmail.com> wrote:
> > As you know, i am new guy about the Mahout. suppose i have one model
> > trained in WEKA using distinct classifiers, if the Mahout have some
> > port to import the model, and using the model in the up-coming
> > process, it will be very cool.
>
> Could you please explain exactly which models you would like to import
> and why?
>
> Assuming we are talking about naive bayes: What is really expensive
> about it is training the classifier. I wonder why you would want to do
> that within Weka.
>
> With most classification algorithms I am familiar with, training is
> expensive, but application to new instances is cheap. That is why
> currently I do not really understand, why you would want to run the
> training in Weka and use the model in Mahout. However, I could imagine
> use cases where you might want to train the model with Mahout and use
> it as part of a processing chain within Weka.
>
>
> Isabel
>
Re: [jira] Commented: (MAHOUT-18) Embrace interoperability with
other softwares
Posted by Isabel Drost <is...@apache.org>.
On Tue Andrew Wang <an...@gmail.com> wrote:
> As you know, i am new guy about the Mahout. suppose i have one model
> trained in WEKA using distinct classifiers, if the Mahout have some
> port to import the model, and using the model in the up-coming
> process, it will be very cool.
Could you please explain exactly which models you would like to import
and why?
Assuming we are talking about naive bayes: What is really expensive
about it is training the classifier. I wonder why you would want to do
that within Weka.
With most classification algorithms I am familiar with, training is
expensive, but application to new instances is cheap. That is why
currently I do not really understand, why you would want to run the
training in Weka and use the model in Mahout. However, I could imagine
use cases where you might want to train the model with Mahout and use
it as part of a processing chain within Weka.
Isabel
Re: [jira] Commented: (MAHOUT-18) Embrace interoperability with other
softwares
Posted by Andrew Wang <an...@gmail.com>.
As you know, i am new guy about the Mahout. suppose i have one model trained
in WEKA using distinct classifiers, if the Mahout have some port to import
the model, and using the model in the up-coming process, it will be very
cool.
Sorry to make you confuse!
On Tue, Nov 17, 2009 at 5:20 PM, Ted Dunning <te...@gmail.com> wrote:
> Can you say more about what you really need?
>
>
>
> On Tue, Nov 17, 2009 at 1:16 AM, Andrew Wang <andrew.wang.1900@gmail.com
> >wrote:
>
> > Sorry to disturb you guys! I am data mining programmer using WEKA(
> > http://www.cs.waikato.ac.nz/ml/weka/) for several years. This funcation
> > may
> > be very useful for us who'd like to import the trained model in WEKA into
> > Mahout.
> > Look forward to it.
> >
>
>
>
> --
> Ted Dunning, CTO
> DeepDyve
>
Re: [jira] Commented: (MAHOUT-18) Embrace interoperability with other
softwares
Posted by Ted Dunning <te...@gmail.com>.
Can you say more about what you really need?
On Tue, Nov 17, 2009 at 1:16 AM, Andrew Wang <an...@gmail.com>wrote:
> Sorry to disturb you guys! I am data mining programmer using WEKA(
> http://www.cs.waikato.ac.nz/ml/weka/) for several years. This funcation
> may
> be very useful for us who'd like to import the trained model in WEKA into
> Mahout.
> Look forward to it.
>
--
Ted Dunning, CTO
DeepDyve