I found that only item-based recommender is implemented in version-0.6.
When I want to use the user-based recommender, all I need to do is to
transpose the input, i.e., <uid, itemid, rating> ---> <itemid, uid,
rating>，right? Is there any other differences between item-based and
user-based approach, which is not noticed by me?

--
Wei Feng

Both of these have been implemented since before Mahout. Can you clarify?
Don't understand at the moment.
On May 9, 2012 3:42 PM, "冯伟" <[EMAIL PROTECTED]> wrote:

> I found that only item-based recommender is implemented in version-0.6.
> When I want to use the user-based recommender, all I need to do is to
> transpose the input, i.e., <uid, itemid, rating> ---> <itemid, uid,
> rating>，right? Is there any other differences between item-based and
> user-based approach, which is not noticed by me?
>
> --
> Wei Feng
>

Sorry for my poor English. My question is like this: When I use
"org.apache.mahout.cf.taste.hadoop.item.RecommenderJob", my input file is
in the format:
"uid, itemid, rating"
"uid, itemid, rating"
"uid, itemid, rating"
According to the above input,
org.apache.mahout.cf.taste.hadoop.item.RecommenderJob will give item-based
recommendations. When I exchange the postions of uid and itemid, the input
file would be
"itemid, uid, rating"
"itemid, uid, rating"
"itemid, uid, rating"
According to the above input,
org.apache.mahout.cf.taste.hadoop.item.RecommenderJob would generate
user-based recommendation. Is that right?

2012/5/9 Sean Owen <[EMAIL PROTECTED]>

> Both of these have been implemented since before Mahout. Can you clarify?
> Don't understand at the moment.
> On May 9, 2012 3:42 PM, "冯伟" <[EMAIL PROTECTED]> wrote:
>
> > I found that only item-based recommender is implemented in version-0.6.
> > When I want to use the user-based recommender, all I need to do is to
> > transpose the input, i.e., <uid, itemid, rating> ---> <itemid, uid,
> > rating>，right? Is there any other differences between item-based and
> > user-based approach, which is not noticed by me?
> >
> > --
> > Wei Feng
> >
>

--
Wei Feng

Ah, distributed implementation. no there is no such implementation now, but
see: https://issues.apache.org/jira/browse/MAHOUT-1004

On Wed, May 9, 2012 at 4:19 PM, 冯伟 <[EMAIL PROTECTED]> wrote:

> Sorry for my poor English. My question is like this: When I use
> "org.apache.mahout.cf.taste.hadoop.item.RecommenderJob", my input file is
> in the format:
> "uid, itemid, rating"
> "uid, itemid, rating"
> "uid, itemid, rating"
> According to the above input,
> org.apache.mahout.cf.taste.hadoop.item.RecommenderJob will give item-based
> recommendations. When I exchange the postions of uid and itemid, the input
> file would be
> "itemid, uid, rating"
> "itemid, uid, rating"
> "itemid, uid, rating"
> According to the above input,
> org.apache.mahout.cf.taste.hadoop.item.RecommenderJob would generate
> user-based recommendation. Is that right?
>
> 2012/5/9 Sean Owen <[EMAIL PROTECTED]>
>
> > Both of these have been implemented since before Mahout. Can you clarify?
> > Don't understand at the moment.
> > On May 9, 2012 3:42 PM, "冯伟" <[EMAIL PROTECTED]> wrote:
> >
> > > I found that only item-based recommender is implemented in version-0.6.
> > > When I want to use the user-based recommender, all I need to do is to
> > > transpose the input, i.e., <uid, itemid, rating> ---> <itemid, uid,
> > > rating>，right? Is there any other differences between item-based and
> > > user-based approach, which is not noticed by me?
> > >
> > > --
> > > Wei Feng
> > >
> >
>
>
>
> --
> Wei Feng
>

If you transpose the input, you would recommend users to items, not
items to users, thats the problem.

--sebastian

On 09.05.2012 16:41, 锟斤拷伟 wrote:
> I found that only item-based recommender is implemented in version-0.6.
> When I want to use the user-based recommender, all I need to do is to
> transpose the input, i.e., <uid, itemid, rating> ---> <itemid, uid,
> rating>锟斤拷right? Is there any other differences between item-based and
> user-based approach, which is not noticed by me?
>

Aha! Thanks! I was only considering about computing the pair-wise
similarity. Now I got it.

2012/5/9 Sebastian Schelter <[EMAIL PROTECTED]>

> If you transpose the input, you would recommend users to items, not
> items to users, thats the problem.
>
> --sebastian
>
> On 09.05.2012 16:41, 冯伟 wrote:
> > I found that only item-based recommender is implemented in version-0.6.
> > When I want to use the user-based recommender, all I need to do is to
> > transpose the input, i.e., <uid, itemid, rating> ---> <itemid, uid,
> > rating>，right? Is there any other differences between item-based and
> > user-based approach, which is not noticed by me?
> >
>
>
--
Wei Feng