|
Is there literature on learning from purchase history? In other words, series of market baskets over time. |
|
Hi, there was a paper at this year's WWW conference: Steffen Rendle, Christoph Freudenthaler, Lars Schmidt-Thieme (2010): Factorizing Personalized Markov Chains for Next-Basket Recommendation, in Proceedings of the 19th International World Wide Web Conference (WWW 2010) This work takes the preceding basket into account. Or do you want more general literature on item recommendation from implicit, positive-only feedback (say, 'view' or 'purchase' events)? Best regards, Z. |
|
Eduardo, considering positive-only feedback, two nice papers to start with are:
There is also (beware! advertisement!)
Of course, there are plenty more publications on the topic, just follow the references/keywords in the papers I mention, and you should find most of them. Best regards, Z. |
|
Hi Z, I would appreciate information on positive-only feedback! Regards, eduardofv What does this even mean?
(Jul 01 '10 at 18:28)
Joseph Turian ♦♦
When all you have is information on positive behaviour (as: I like this and this) as opposed to negative and positive (as in I like this and I don't like this). Big difference if you are trying to discriminate preference, since a non-negative feedback can mean both I don't like or I didn't saw the item.
(Jul 02 '10 at 11:03)
eduardofv
Gotcha. This belongs as a comment on @zeno's answer, tho, not as an answer by itself.
(Jul 03 '10 at 00:42)
Joseph Turian ♦♦
|
