mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Grant Ingersoll <gsing...@apache.org>
Subject Re: [Taste] Sanity Check and Questions
Date Fri, 19 Jun 2009 13:45:15 GMT

On Jun 19, 2009, at 8:47 AM, Sean Owen wrote:

> On Fri, Jun 19, 2009 at 8:40 AM, Grant  
> Ingersoll<gsingers@apache.org> wrote:
>> I don't know, "Programming Collective Intelligence" (page 17) seems  
>> to
>> suggest transposing as being pretty useful for product-product  
>> suggestions.
>>  I suppose, however, that it really is just getting the  
>> neighborhood then
>> and not going the extra step of getting items (er users in this  
>> case).
>>
>> The main cost is in having a second model in memory, I suppose.
>
> Yes, I mean, it works. I think they were showing this approach for
> simplicity and brevity. By turning a user-based recommender on its
> side you kind of get an item-based recommender, which is computing
> item-item similarities under the hood, and indeed that is the basis of
> what you want. My suggestion a) just goes straight to that.

Yeah, Taste's TopItems is the equiv of the topMatches() method in the  
book, it seems.

Also, as opposed to calling TopItems directly, I could just call the  
mostSimilarItems method on GenericBasedItemRecommender, right?  That  
way I can rely on the recommender for the Estimator, etc. and the API  
seems cleaner.

Mime
View raw message