mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sean Owen <sro...@gmail.com>
Subject Re: Improve Recommendations
Date Thu, 10 May 2012 16:10:03 GMT
You could make an ItemSimilarity based on the tags. Maybe you use the
Tanimoto coefficient to make a simple similarity metric based on their
presence or absence. Then you use an item-based recommender.

On Thu, May 10, 2012 at 5:07 PM, Jahangir Mohammed
<md.jahangir27@gmail.com> wrote:
> Thanks Sean for response. Let say, I have tags associated to the items
> where items are more like an e-commerce(electronics, less than 20$, best
> buy product,..so on) and I have some users demographic information. So, I
> was thinking that I can write UserSimilarity and ItemSimilarity based on
> these.
>
> On Thu, May 10, 2012 at 11:52 AM, Sean Owen <srowen@gmail.com> wrote:
>
>> The best, or perhaps only, way to integrate such information is to
>> implement your own version of UserSimilarity or ItemSimilarity and
>> then use a user-based or item-based recommender. You can implement
>> whatever similarity rule you think is best according to your metadata.
>> There's not a lot to be said unless you can say more about exactly
>> what metadata you have.
>>
>> On Thu, May 10, 2012 at 4:31 PM, Jahangir Mohammed
>> <md.jahangir27@gmail.com> wrote:
>> > I followed very much the example as given in the documentation and it's
>> > working as expected so far.
>> >
>> > Have a question on how I can use some of the metadata I have for users
>> and
>> > items to still improve the recommendations. It looks obvious how to do
>> it,
>> > but still want to get a heads up from people who have been using taste
>> for
>> > a while. So, is my assumption correct that I have to implement
>> > UserSimilarity and ItemSimilarity interfaces?
>> >
>> > Thanks in advance for any suggestions.
>> >
>> > Thanks,
>> > Jahangir.
>>

Mime
View raw message