You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@mahout.apache.org by WoodJustin <hu...@hotmail.com> on 2010/07/13 17:43:37 UTC
Online Recommendation
Hi all,
I am reading the <Mahout in Action> and run the code of user-based recommendation based on 10 million rating data set. It seems extremly slow to calculate the recommended items. I am curious if I want to do the online recommendation, how could I do?
Thank you for your attention.
---Tony
_________________________________________________________________
约会说不清地方？来试试微软地图最新msn互动功能！
http://ditu.live.com/?form=TL&swm=1
Re: Re: Online Recommendation
Posted by Sean Owen <sr...@gmail.com>.
10 requests per second sounds reasonable. You might wish to test but I
believe one processor core on one server can handle that.
2010/7/13 Young <wo...@126.com>:
> Thank you, Sean. That makes sense. It uses 1 second to generete recommendations. After intiating the recommender, the recommend() will be very fast. I just wondering if there are multiple requests per second, say 10 requests, does this work fine?
>
Re:Re: Online Recommendation
Posted by Young <wo...@126.com>.
Thank you, Sean. That makes sense. It uses 1 second to generete recommendations. After intiating the recommender, the recommend() will be very fast. I just wondering if there are multiple requests per second, say 10 requests, does this work fine?
>OK, but these are not log messages from producing recommendations.
>This shows loading the data into memory the first time. It may take a
>little time, but, producing recommendations after that should be very
>fast.
>
>2010/7/13 WoodJustin <hu...@hotmail.com>:
>>
>> i did not modify anything in the Mahout in Action. First new GroupLensDataModel, then PearsonCorrelationSimilarity and Neighborhood, and last generate the recommendation. And this the result
>>
>> 12:12:59
>> Info: Creating FileDataModel
>>
>> 12:13:01
>> Info: Reading file info...
>> ...
>> 12:14:59
>> Info: Processed 10000000 lines
>>
>> 12:14:59
>> Info: Read lines: 10000054
>>
>> 12:15:03
>> Info: Processed 69878 users
>>
>>
>>
>> _________________________________________________________________
>> MSN十年回馈，每位用户可免费获得价值25元的卡巴斯基反病毒软件2010激活码，快来领取！
>> http://kaba.msn.com.cn/?k=1
Re: Online Recommendation
Posted by Sean Owen <sr...@gmail.com>.
OK, but these are not log messages from producing recommendations.
This shows loading the data into memory the first time. It may take a
little time, but, producing recommendations after that should be very
fast.
2010/7/13 WoodJustin <hu...@hotmail.com>:
>
> i did not modify anything in the Mahout in Action. First new GroupLensDataModel, then PearsonCorrelationSimilarity and Neighborhood, and last generate the recommendation. And this the result
>
> 12:12:59
> Info: Creating FileDataModel
>
> 12:13:01
> Info: Reading file info...
> ...
> 12:14:59
> Info: Processed 10000000 lines
>
> 12:14:59
> Info: Read lines: 10000054
>
> 12:15:03
> Info: Processed 69878 users
>
>
>
> _________________________________________________________________
> MSN十年回馈，每位用户可免费获得价值25元的卡巴斯基反病毒软件2010激活码，快来领取！
> http://kaba.msn.com.cn/?k=1
RE: Online Recommendation
Posted by WoodJustin <hu...@hotmail.com>.
i did not modify anything in the Mahout in Action. First new GroupLensDataModel, then PearsonCorrelationSimilarity and Neighborhood, and last generate the recommendation. And this the result
12:12:59
Info: Creating FileDataModel
12:13:01
Info: Reading file info...
...
12:14:59
Info: Processed 10000000 lines
12:14:59
Info: Read lines: 10000054
12:15:03
Info: Processed 69878 users
_________________________________________________________________
MSN十年回馈，每位用户可免费获得价值25元的卡巴斯基反病毒软件2010激活码，快来领取！
http://kaba.msn.com.cn/?k=1
Re: Online Recommendation
Posted by Sean Owen <sr...@gmail.com>.
10 million data points is not large, and should not be "slow".
Recommendations should take less than 100ms with normal algorithms and
data sets. What are you seeing?
There are many, many ways to make a recommender run very slowly, and a
few ways to do it right.
Without any details, it's not possible to comment more. What data,
what algorithm, etc?
2010/7/13 WoodJustin <hu...@hotmail.com>:
>
> Hi all,
>
>
>
> I am reading the <Mahout in Action> and run the code of user-based recommendation based on 10 million rating data set. It seems extremly slow to calculate the recommended items. I am curious if I want to do the online recommendation, how could I do?
>
>
>
> Thank you for your attention.
>
>
>
> ---Tony
>
> _________________________________________________________________
> 约会说不清地方？来试试微软地图最新msn互动功能！
> http://ditu.live.com/?form=TL&swm=1