You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mahout.apache.org by Grant Ingersoll <gs...@apache.org> on 2008/02/12 15:15:32 UTC
Karl's Lucene donations
Hey Karl,
Does it make sense to try to bring some of your Lucene donations of ML
things over to Mahout, i.e.: https://issues.apache.org/jira/browse/LUCENE-1039
, https://issues.apache.org/jira/browse/LUCENE-1025
and maybe even https://issues.apache.org/jira/browse/LUCENE-625 and
maybe https://issues.apache.org/jira/browse/LUCENE-826, others????
At any rate, I may look at using 1039 as a seed for the Bayesian
classifier.
-Grant
Re: Karl's Lucene donations
Posted by Grant Ingersoll <gs...@apache.org>.
On Feb 12, 2008, at 10:30 AM, Karl Wettin wrote:
>
> 12 feb 2008 kl. 15.15 skrev Grant Ingersoll:
>
>> Hey Karl,
>>
>> Does it make sense to try to bring some of your Lucene donations of
>> ML things over to Mahout
>
> Perhaps.
>
> 626 could do with a bit of M/R at training, but I'm not sure it fits
> this project. Also, it is quite unscientific. But I would rather see
> it replace the current Lucene spellcheck, as it can do that.
>
>> At any rate, I may look at using 1039 as a seed for the Bayesian
>> classifier.
>
> This might be the best way to use the code. But make sure you read
> the comments in the Jira! I've sent them a mail a few months ago,
> but never heard anything.
I did. I don't know that I even have to look at it, but is always
good to have other opinions. Still have some more back reading to do
to see if there are any things I should be thinking about. Then, of
course, is trying to find some time slices to actually do it. But,
heh, who needs sleep?
-Grant
Re: Karl's Lucene donations
Posted by Karl Wettin <ka...@gmail.com>.
12 feb 2008 kl. 15.15 skrev Grant Ingersoll:
> Hey Karl,
>
> Does it make sense to try to bring some of your Lucene donations of
> ML things over to Mahout
Perhaps.
626 could do with a bit of M/R at training, but I'm not sure it fits
this project. Also, it is quite unscientific. But I would rather see
it replace the current Lucene spellcheck, as it can do that.
> At any rate, I may look at using 1039 as a seed for the Bayesian
> classifier.
This might be the best way to use the code. But make sure you read the
comments in the Jira! I've sent them a mail a few months ago, but
never heard anything.
My first real ML contribtion to Mahout will probably similar to 1025,
heirarcial clusterer.
karl