I was finally able to get things running. I had completely overlooked
the similarity algorithm parameter, and once I set that, it worked
great! It'd be nice if mahout complained immediately when a required
param is missing instead of when the param is needed, especially in
multi-phase/map-reduce jobs.

- Matt

On Sat, Aug 4, 2012 at 5:11 PM, Matt Mitchell <[email protected]> wrote:
> I'm attempting to run the
> org.apache.mahout.cf.taste.hadoop.similarity.item.ItemSimilarityJob on
> AWS EMR.
>
> I can see that things are working by looking at the task logs.
> However, after it runs for about 10 minutes, it dies. The only log
> file is stdout, and it's empty.
>
> Does this look right -- using the ruby client:
>
> ./elastic-mapreduce -j JOB_ID --jar
> s3n://mm.lib/mahout-core-0.6-job.jar --main-class
> org.apache.mahout.cf.taste.hadoop.similarity.item.ItemSimilarityJob
> --arg --input --arg s3n://mm.input-data/data.csv --arg --output --arg
> s3n://mm.output-data/ --arg --tempDir --arg tempDir4 --access-id
> ACCESS_KEY --private-key PRIVATE_KEY
>
> One question... should the S3 output directory already exist?
>
> - Matt
>
> On Sat, Aug 4, 2012 at 3:18 PM, Matt Mitchell <[email protected]> wrote:
>> Thanks :) Of course, I found this as soon as I posted!
>>
>> https://cwiki.apache.org/MAHOUT/mahout-on-elastic-mapreduce.html
>>
>> - Matt
>>
>> On Sat, Aug 4, 2012 at 2:34 PM, Sebastian Schelter <[email protected]> wrote:
>>> Its pretty simple, upload the mahout jar and your data to S3 and click
>>> together a custom mapreduce step pointing to the ItemSimilarityJob class
>>> Am 04.08.2012 20:29 schrieb "Matt Mitchell" <[email protected]>:
>>>
>>>> Hi,
>>>>
>>>> I'm digging around trying to find info on running mahout on AWS's
>>>> Elastic Map Reduce. Anyone know of a step-by-step article/tutorial?
>>>> I'm interested in running "itemsimilarity", "recommenditembased" and
>>>> "recommendfactorized".
>>>>
>>>> Thanks!
>>>>
>>>> - Matt
>>>>

Reply via email to