Closed ShaneColeman closed 6 years ago
Hi,
They are some hyper-parameters of the FISM algorithm.
On 14 May 2018 at 06:17, Shane Coleman notifications@github.com wrote:
Hello,
I have a question which relates to both FISMrmse and FISMauc Recommender Systems.
Within the Algorithm List for both Recommenders, may I ask, what do the following lines for both Recommenders represent:
FISMauc Recommender rec.fismauc.rho=2 rec.fismauc.alpha=1.5
FISMrmse Recommender rec.fismrmse.rho=1 rec.fismrmse.alpha=1.5
I have read the associated paper: Kabbur S, Ning X, Karypis G. Fism: factored item similarity models for top-n recommender systems[C]//KDD. 2013, and I am just looking for clarification to fully understand what these lines mean in relation to the paper?
If there is any other information you require, please do not hesitate to ask. Your help and feedback would be greatly appreciated.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/guoguibing/librec/issues/257, or mute the thread https://github.com/notifications/unsubscribe-auth/AEN9viO10_SCIOi5MpuHCtcao-tPHBAyks5tyVm3gaJpZM4T9hZM .
Hi allenjack,
Thank you very much for your response.
If possible, would you be able to tell me (in relation to the FISM algorithms (FISMrmse & FISMauc) hyper-parameters) what .rho and .alpha represents. I have read the paper associted with the FISM recommender system, and sadly cannot find any mention to these hyper-parameters. Would it be possible that these hyper-parameters are represented or known as something else in the paper?
Again, thank you very much for your help.
Let me have a look, and reach you out later.
On 16 May 2018 at 11:35, Shane Coleman notifications@github.com wrote:
Hi allenjack,
Thank you very much for your response.
If possible, would you be able to tell me (in relation to the FISM algorithms (FISMrmse & FISMauc) hyper-parameters) what .rho and .alpha represents. I have read the paper associted with the FISM recommender system, and sadly cannot find any mention to these hyper-parameters. Would it be possible that these hyper-parameters are represented or known as something else in the paper?
Again, thank you very much for your help.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/guoguibing/librec/issues/257#issuecomment-389563825, or mute the thread https://github.com/notifications/unsubscribe-auth/AEN9vrqdYveLj0e81AQ1r7iRJ7HaRO8Gks5tzEcsgaJpZM4T9hZM .
Thank you allenjack, I really appreciate it.
Hi,
For alpha:
For rho:
Hi, allenjack,
Thank you very much taking the time to help me answer this question. I really appreciate it.
If possible, could you tell me what .rho stands for in relation to the hyper-parameter?
Again, thanks for your help
Hello,
I have a question which relates to both FISMrmse and FISMauc Recommender Systems.
Within the Algorithm List for both Recommenders, may I ask, what do the following lines for both Recommenders represent:
FISMauc Recommender rec.fismauc.rho=2 rec.fismauc.alpha=1.5
FISMrmse Recommender rec.fismrmse.rho=1 rec.fismrmse.alpha=1.5
I have read the associated paper: Kabbur S, Ning X, Karypis G. Fism: factored item similarity models for top-n recommender systems[C]//KDD. 2013, and I am just looking for clarification to fully understand what these lines mean in relation to the paper?
If there is any other information you require, please do not hesitate to ask. Your help and feedback would be greatly appreciated.