Re: set-similarity in mahout

[Ted Dunning <td...@deepdyve.com>]

Rares, Chenli,


I don't have the specific classes at hand but here are some pointers to
related items ...

- the frequent item-set stuff is related (
http://tdunning.blogspot.com/2010/04/hadoop-user-group-aka-mahout-users.htmland
https://cwiki.apache.org/MAHOUT/parallel-frequent-pattern-mining.html )

- the recommendation system has a cooccurrence counter (recently merged with
similar code:
http://mail-archives.apache.org/mod_mbox/lucene-mahout-dev/201002.mbox/%3C262964957.461161266929907859.JavaMail.jira@brutus.apache.org%3E
)

- the large scale SVD code includes an efficient A' A multiplier (some of
the discussion is here: https://issues.apache.org/jira/browse/MAHOUT-180 but
this is very old and only useful for beginnings of pointers)


Other Mahouts,

This question came out of my slightly less than gracious questioning to Chen
Li and Rares after their talk at the Hadoop Summit.  This is their very
gracious followup which I have taken the liberty of forwarding to the list
to see if anybody can quickly amplify the comments above.

Does anybody have more specific pointers?


On Thu, Jul 1, 2010 at 10:12 AM, Chen Li <ch...@ics.uci.edu> wrote:

> Ted,
>
> I want to add my thanks to you for your questions and interests in our
> work.  We will appreciate it if you can provide us information about
> the related module in Mahout.
>
> Chen
>
> On Thu, Jul 1, 2010 at 9:35 AM, Rares Vernica <ra...@ics.uci.edu> wrote:
> > Hello Ted,
> >
> > It was very nice meeting you at the Hadoop Summit. Thanks for your
> > feedback on our set-similarity join work. To follow up, could you point
> > us to the algorithm/module that does the equivalent of set-similarity
> > join in the mahout project?
> >
> > Thank you,
> > Rares Vernica
> > UC Irvine
> >
> >
>

Re: set-similarity in mahout

[Chen Li <ch...@ics.uci.edu>]

Ted,

Thanks for the links.  We will look into them.  For other Mahout
developers, our work and source code on set-similarity joins using
Hadoop are available at:
http://asterix.ics.uci.edu/fuzzyjoin-mapreduce/ .  Comments are
welcome.

Chen

On Thu, Jul 1, 2010 at 11:10 AM, Ted Dunning <td...@deepdyve.com> wrote:
> Rares, Chenli,
>
> I don't have the specific classes at hand but here are some pointers to
> related items ...
> - the frequent item-set stuff is related
> (http://tdunning.blogspot.com/2010/04/hadoop-user-group-aka-mahout-users.html
> and https://cwiki.apache.org/MAHOUT/parallel-frequent-pattern-mining.html )
> - the recommendation system has a cooccurrence counter (recently merged with
> similar
> code: http://mail-archives.apache.org/mod_mbox/lucene-mahout-dev/201002.mbox/%3C262964957.461161266929907859.JavaMail.jira@brutus.apache.org%3E)
> - the large scale SVD code includes an efficient A' A multiplier (some of
> the discussion is here: https://issues.apache.org/jira/browse/MAHOUT-180 but
> this is very old and only useful for beginnings of pointers)
>
> Other Mahouts,
>
> This question came out of my slightly less than gracious questioning to Chen
> Li and Rares after their talk at the Hadoop Summit.  This is their very
> gracious followup which I have taken the liberty of forwarding to the list
> to see if anybody can quickly amplify the comments above.
> Does anybody have more specific pointers?
>
> On Thu, Jul 1, 2010 at 10:12 AM, Chen Li <ch...@ics.uci.edu> wrote:
>>
>> Ted,
>>
>> I want to add my thanks to you for your questions and interests in our
>> work.  We will appreciate it if you can provide us information about
>> the related module in Mahout.
>>
>> Chen
>>
>> On Thu, Jul 1, 2010 at 9:35 AM, Rares Vernica <ra...@ics.uci.edu> wrote:
>> > Hello Ted,
>> >
>> > It was very nice meeting you at the Hadoop Summit. Thanks for your
>> > feedback on our set-similarity join work. To follow up, could you point
>> > us to the algorithm/module that does the equivalent of set-similarity
>> > join in the mahout project?
>> >
>> > Thank you,
>> > Rares Vernica
>> > UC Irvine
>> >
>> >
>
>