You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@mahout.apache.org by Anatoliy Kats <a....@rambler-co.ru> on 2011/11/29 10:32:39 UTC
Time-based preferences for recommendation
Hi,
There was a conversation some time ago about incorporating time
dependency for preferences:
http://thread.gmane.org/gmane.comp.apache.mahout.user/2951
Has there been any more discussion about this? Has anything been
checked into Mahout? Is anyone working on it? I might be able to pitch in.
Thanks,
Anatoliy
Re: Time-based preferences for recommendation
Posted by Anatoliy Kats <a....@rambler-co.ru>.
Ah wow, thanks for that list. I will take a look at some of those
within the next couple of weeks.
On 11/30/2011 02:12 AM, Christoph Hermann wrote:
> Am Dienstag, 29. November 2011, 10:32:39 schrieb Anatoliy Kats:
>
> Hello,
>
>> There was a conversation some time ago about incorporating time
>> dependency for preferences:
>> http://thread.gmane.org/gmane.comp.apache.mahout.user/2951
>>
>> Has there been any more discussion about this? Has anything been
>> checked into Mahout? Is anyone working on it? I might be able to pitch
>> in.
> i did some more work on this some time ago. But i was using data from a course
> system which probably differs a lot from usual shop recommendations.
>
> Some of my work is described in this paper:
> Time-Based Recommendations for Lecture Materials
> In Proceedings of ED-MEDIA 2010 Toronto, Canada, Jul. 2010
> http://algo.informatik.uni-freiburg.de/mitarbeiter/hermann/files/aace-ed-
> media-2010-word-final.pdf
> You should also have a look at these publications:
>
> Iain Campbell and C. J. van Rijsbergen. The ostensive model of developing
> information needs. In Peter Ingwersen and Niels Ole Pors, editors, CoLIS’06:
> Proceedings of the 6th International Conference on Conceptions of Library and
> Information Sciences, Copenhagen, Denmark, pages 251–268. The Royal School of
> Librarianship, 10 1996.
>
> Iain Campbell. Supporting information needs by ostensive definition in an
> adaptive information space. In Ian Ruthven, editor, MIRO’95: Proceedings of
> the Final Workshop on Multimedia Information Retrieval, Glasgow, Scotland, UK,
> Workshops in Computing. BCS, 9 1995.
>
> Iain Campbell. Interactive Evaluation of the Ostensive Model Using a New Test
> Collection of Images with Multiple Relevance Assessments. Information
> Retrieval, 2(1): 85–112, 2000.
>
> Collaborative Filtering with Temporal Dynamics
> KDD'09 from Yehuda Koren, Yahoo! Research, Haifa, Israel
>
> Factorizing Personalized Markov Chains for Next-Basket Recommendation
> WWW 2010 from Steffen Rendle, Christoph Freudenthaler, Lars Schmidt-Thieme
>
> Yi Ding, Xue Li, Maria E. Orlowska (2006): /Recency-based collaborative
> filtering/ CRPITS'49: Proceedings of the 49th conference on Database
> technologies 2006, ACM Press, pp. 99--107.
>
> Sinan Zhan, Fengrong Gao, Chunxiao Xing, Lizhu Zhou (2006): /Addressing
> Concept Drift problem in collaborative filtering systems/ Proceedings of ECAI
> 2006 Workshop on Recommender Systems, Riva del Garda, Italy.
>
> Yi Ding, Xue Li (2005): Time weight collaborative filtering CIKM '05:
> Proceedings of the 14th ACM international conference on Information and
> knowledge management, ACM Press, pp. 485--492.
>
> Tiffany Ya Tang, Pinata Winoto, Keith C. C. Chan (2003): /On the Temporal
> Analysis for Improved Hybrid Recommendations/ WI '03: Proceedings of the
> IEEE/WIC International Conference on Web Intelligence, IEEE Computer Society.
>
> regards
> Christoph
Re: Time-based preferences for recommendation
Posted by Christoph Hermann <ch...@guschtel.de>.
Am Dienstag, 29. November 2011, 10:32:39 schrieb Anatoliy Kats:
Hello,
> There was a conversation some time ago about incorporating time
> dependency for preferences:
> http://thread.gmane.org/gmane.comp.apache.mahout.user/2951
>
> Has there been any more discussion about this? Has anything been
> checked into Mahout? Is anyone working on it? I might be able to pitch
> in.
i did some more work on this some time ago. But i was using data from a course
system which probably differs a lot from usual shop recommendations.
Some of my work is described in this paper:
Time-Based Recommendations for Lecture Materials
In Proceedings of ED-MEDIA 2010 Toronto, Canada, Jul. 2010
http://algo.informatik.uni-freiburg.de/mitarbeiter/hermann/files/aace-ed-
media-2010-word-final.pdf
You should also have a look at these publications:
Iain Campbell and C. J. van Rijsbergen. The ostensive model of developing
information needs. In Peter Ingwersen and Niels Ole Pors, editors, CoLIS’06:
Proceedings of the 6th International Conference on Conceptions of Library and
Information Sciences, Copenhagen, Denmark, pages 251–268. The Royal School of
Librarianship, 10 1996.
Iain Campbell. Supporting information needs by ostensive definition in an
adaptive information space. In Ian Ruthven, editor, MIRO’95: Proceedings of
the Final Workshop on Multimedia Information Retrieval, Glasgow, Scotland, UK,
Workshops in Computing. BCS, 9 1995.
Iain Campbell. Interactive Evaluation of the Ostensive Model Using a New Test
Collection of Images with Multiple Relevance Assessments. Information
Retrieval, 2(1): 85–112, 2000.
Collaborative Filtering with Temporal Dynamics
KDD'09 from Yehuda Koren, Yahoo! Research, Haifa, Israel
Factorizing Personalized Markov Chains for Next-Basket Recommendation
WWW 2010 from Steffen Rendle, Christoph Freudenthaler, Lars Schmidt-Thieme
Yi Ding, Xue Li, Maria E. Orlowska (2006): /Recency-based collaborative
filtering/ CRPITS'49: Proceedings of the 49th conference on Database
technologies 2006, ACM Press, pp. 99--107.
Sinan Zhan, Fengrong Gao, Chunxiao Xing, Lizhu Zhou (2006): /Addressing
Concept Drift problem in collaborative filtering systems/ Proceedings of ECAI
2006 Workshop on Recommender Systems, Riva del Garda, Italy.
Yi Ding, Xue Li (2005): Time weight collaborative filtering CIKM '05:
Proceedings of the 14th ACM international conference on Information and
knowledge management, ACM Press, pp. 485--492.
Tiffany Ya Tang, Pinata Winoto, Keith C. C. Chan (2003): /On the Temporal
Analysis for Improved Hybrid Recommendations/ WI '03: Proceedings of the
IEEE/WIC International Conference on Web Intelligence, IEEE Computer Society.
regards
Christoph
Re: Time-based preferences for recommendation
Posted by Anatoliy Kats <a....@rambler-co.ru>.
Hi Manuel,
Thank you for the reference. I am just testing the waters for now,
trying to find out what's available. I should have a usecase in a
couple of weeks. I'll reread what's said here then, and continue the
thread.
Cheers,
Anatoliy
On 11/29/2011 03:21 PM, Manuel Blechschmidt wrote:
> Hello Anatoliy,
>
> On 29.11.2011, at 10:32, Anatoliy Kats wrote:
>
>> Hi,
>>
>> There was a conversation some time ago about incorporating time dependency for preferences: http://thread.gmane.org/gmane.comp.apache.mahout.user/2951
>>
>> Has there been any more discussion about this? Has anything been checked into Mahout? Is anyone working on it? I might be able to pitch in.
>
> I am currently working with a data set which has highly seasonal data. Actually it is the sales data of a merchant selling tea and spices.
>
> I benchmarked the different recommenders against it:
> http://thread.gmane.org/gmane.comp.apache.mahout.user/10433
>
> As far as I know there are currently no recommenders that incorporate time or seasons. The DataModel supports it but it isn't used.
>
> I would guess that identifying seasonal patterns could enhance my recommendations a lot.
>
> I scanned the following paper:
> Improving E-Commerce Recommender Systems by the Identification of Seasonal Products
> http://www.aaai.org/Papers/Workshops/2007/WS-07-08/WS07-08-011.pdf
>
> Actually I think that what the paper is doing is not that advanced.
>
> I currently try to identify seasonal products with R. I am playing around with seasonal ARIMA models (http://www.duke.edu/~rnau/seasarim.htm http://cran.r-project.org/web/packages/forecast/forecast.pdf). If I have a working solution with R I might implement it in Mahout.
>
> What is your use case? Do you already have a data set?
>
>> Thanks,
>>
>> Anatoliy
> /Manuel
>
Re: Time-based preferences for recommendation
Posted by Dan Brickley <da...@danbri.org>.
On 29 November 2011 16:11, Ted Dunning <te...@gmail.com> wrote:
> The deanonymization attacks depend on some aspect of the data being related
> to real-world events or products. The attack on the netflix data depended
> on the movies being identified so that ratings could be correlated to
> ratings on other systems.
>
> If you blind product id's and user id's then none of the currently known
> attacks are likely to work.
>
> But I respect the limits you describe. Caution is never such a bad thing
> in these matters.
+1 ... data leaks in expected ways. Random aside: in the Netherlands
currently, Albert Heijn, a major supermarket chain, give away key tags
with a loyalty barcode printed on them. If you enter the code from the
keytag (properly, your own; but maybe you see colleagues and friends
keyring from time to time...) into http://www.ah.nl/appie you get that
person's shopping history ordered by favourites, most recent
purchases. Although as Ted says this only unlocks half the puzzle, ...
Dan
Re: Time-based preferences for recommendation
Posted by Ted Dunning <te...@gmail.com>.
The deanonymization attacks depend on some aspect of the data being related
to real-world events or products. The attack on the netflix data depended
on the movies being identified so that ratings could be correlated to
ratings on other systems.
If you blind product id's and user id's then none of the currently known
attacks are likely to work.
But I respect the limits you describe. Caution is never such a bad thing
in these matters.
On Tue, Nov 29, 2011 at 6:50 AM, Manuel Blechschmidt <
Manuel.Blechschmidt@gmx.de> wrote:
> The problem is that there are quite reliable ways to deanonymize data in a
> reliable way [1]. Further this is also used [2].
>
Re: Time-based preferences for recommendation
Posted by Manuel Blechschmidt <Ma...@gmx.de>.
Hi Ted,
I agree with you. I would love to release it.
Unfortunately it is not my data therefore I can not just release it to public not even anonymized. If someone is willing to contribute new algorithms I can release anonymized data sets on a personal basis.
The problem is that there are quite reliable ways to deanonymize data in a reliable way [1]. Further this is also used [2].
Germany is a lot more restricted about privacy laws.
So if someone is interested in using my dataset send me an email.
/Manuel
[1] Narayanan, Arvind ; Shmatikov, Vitaly: Robust De-anonymization of Large Sparse Datasets. In: Proceedings of the 2008 IEEE Symposium on Security and Privacy. Washington, DC, USA : IEEE Computer Society, 2008. – ISBN 978–0– 7695–3168–7, 111–125
[2] Barbaro, Michael ; Jr., Tom Z.: A Face Is Exposed for AOL Searcher No. 4417749. http://www.nytimes.com/2006/08/09/technology/09aol.html?_r=1. Version: August 2006, Checked: 2011-03-09
[3] http://www.wired.com/threatlevel/2009/12/netflix-privacy-lawsuit/
On 29.11.2011, at 15:31, Ted Dunning wrote:
> Manuel,
>
> If you can blind your data sufficiently to release it publicly, it would
> make it much easier to get others to help with this.
>
> On Tue, Nov 29, 2011 at 3:21 AM, Manuel Blechschmidt <
> Manuel.Blechschmidt@gmx.de> wrote:
>
>> Hello Anatoliy,
>>
>> On 29.11.2011, at 10:32, Anatoliy Kats wrote:
>>
>>> Hi,
>>>
>>> There was a conversation some time ago about incorporating time
>> dependency for preferences:
>> http://thread.gmane.org/gmane.comp.apache.mahout.user/2951
>>>
>>> Has there been any more discussion about this? Has anything been
>> checked into Mahout? Is anyone working on it? I might be able to pitch in.
>>
>>
>> I am currently working with a data set which has highly seasonal data.
>> Actually it is the sales data of a merchant selling tea and spices.
>>
>> I benchmarked the different recommenders against it:
>> http://thread.gmane.org/gmane.comp.apache.mahout.user/10433
>>
>> As far as I know there are currently no recommenders that incorporate time
>> or seasons. The DataModel supports it but it isn't used.
>>
>> I would guess that identifying seasonal patterns could enhance my
>> recommendations a lot.
>>
>> I scanned the following paper:
>> Improving E-Commerce Recommender Systems by the Identification of Seasonal
>> Products
>> http://www.aaai.org/Papers/Workshops/2007/WS-07-08/WS07-08-011.pdf
>>
>> Actually I think that what the paper is doing is not that advanced.
>>
>> I currently try to identify seasonal products with R. I am playing around
>> with seasonal ARIMA models (http://www.duke.edu/~rnau/seasarim.htm
>> http://cran.r-project.org/web/packages/forecast/forecast.pdf). If I have
>> a working solution with R I might implement it in Mahout.
>>
>> What is your use case? Do you already have a data set?
>>
>>>
>>> Thanks,
>>>
>>> Anatoliy
>>
>> /Manuel
>>
>> --
>> Manuel Blechschmidt
>> Dortustr. 57
>> 14467 Potsdam
>> Mobil: 0173/6322621
>> Twitter: http://twitter.com/Manuel_B
>>
>>
--
Manuel Blechschmidt
Dortustr. 57
14467 Potsdam
Mobil: 0173/6322621
Twitter: http://twitter.com/Manuel_B
Re: Time-based preferences for recommendation
Posted by Ted Dunning <te...@gmail.com>.
Manuel,
If you can blind your data sufficiently to release it publicly, it would
make it much easier to get others to help with this.
On Tue, Nov 29, 2011 at 3:21 AM, Manuel Blechschmidt <
Manuel.Blechschmidt@gmx.de> wrote:
> Hello Anatoliy,
>
> On 29.11.2011, at 10:32, Anatoliy Kats wrote:
>
> > Hi,
> >
> > There was a conversation some time ago about incorporating time
> dependency for preferences:
> http://thread.gmane.org/gmane.comp.apache.mahout.user/2951
> >
> > Has there been any more discussion about this? Has anything been
> checked into Mahout? Is anyone working on it? I might be able to pitch in.
>
>
> I am currently working with a data set which has highly seasonal data.
> Actually it is the sales data of a merchant selling tea and spices.
>
> I benchmarked the different recommenders against it:
> http://thread.gmane.org/gmane.comp.apache.mahout.user/10433
>
> As far as I know there are currently no recommenders that incorporate time
> or seasons. The DataModel supports it but it isn't used.
>
> I would guess that identifying seasonal patterns could enhance my
> recommendations a lot.
>
> I scanned the following paper:
> Improving E-Commerce Recommender Systems by the Identification of Seasonal
> Products
> http://www.aaai.org/Papers/Workshops/2007/WS-07-08/WS07-08-011.pdf
>
> Actually I think that what the paper is doing is not that advanced.
>
> I currently try to identify seasonal products with R. I am playing around
> with seasonal ARIMA models (http://www.duke.edu/~rnau/seasarim.htm
> http://cran.r-project.org/web/packages/forecast/forecast.pdf). If I have
> a working solution with R I might implement it in Mahout.
>
> What is your use case? Do you already have a data set?
>
> >
> > Thanks,
> >
> > Anatoliy
>
> /Manuel
>
> --
> Manuel Blechschmidt
> Dortustr. 57
> 14467 Potsdam
> Mobil: 0173/6322621
> Twitter: http://twitter.com/Manuel_B
>
>
Re: Time-based preferences for recommendation
Posted by Manuel Blechschmidt <Ma...@gmx.de>.
Hello Anatoliy,
On 29.11.2011, at 10:32, Anatoliy Kats wrote:
> Hi,
>
> There was a conversation some time ago about incorporating time dependency for preferences: http://thread.gmane.org/gmane.comp.apache.mahout.user/2951
>
> Has there been any more discussion about this? Has anything been checked into Mahout? Is anyone working on it? I might be able to pitch in.
I am currently working with a data set which has highly seasonal data. Actually it is the sales data of a merchant selling tea and spices.
I benchmarked the different recommenders against it:
http://thread.gmane.org/gmane.comp.apache.mahout.user/10433
As far as I know there are currently no recommenders that incorporate time or seasons. The DataModel supports it but it isn't used.
I would guess that identifying seasonal patterns could enhance my recommendations a lot.
I scanned the following paper:
Improving E-Commerce Recommender Systems by the Identification of Seasonal Products
http://www.aaai.org/Papers/Workshops/2007/WS-07-08/WS07-08-011.pdf
Actually I think that what the paper is doing is not that advanced.
I currently try to identify seasonal products with R. I am playing around with seasonal ARIMA models (http://www.duke.edu/~rnau/seasarim.htm http://cran.r-project.org/web/packages/forecast/forecast.pdf). If I have a working solution with R I might implement it in Mahout.
What is your use case? Do you already have a data set?
>
> Thanks,
>
> Anatoliy
/Manuel
--
Manuel Blechschmidt
Dortustr. 57
14467 Potsdam
Mobil: 0173/6322621
Twitter: http://twitter.com/Manuel_B