"Implicit feedback recommendation via implicit-to-explicit ordinal logistic regression mapping"

this article may be useful for your problem..

Look up the basics of web analytics... In implicit rating, we consider things like users' page views, clickthroughs, navigations and so on. For example, the keywords or tags in the pages a user visits more often or spends more time on can be used to build his preference profile

It depends... What kind of implits rating do you have? If there are only web logs on viewed pages, it could be difficult to convert them into different explicit ratings. If there are logs of purchases, the cost of the purchased items could be an interesting rating information. If there are some information about watching TV programs, the time spent, if available, could be interesting.

Dear Frank,

thanks for replying,

I have listening count of music audios..i have seen some papers regarding its conversion but want to know best approach that would give non-biased explicit rating for each listening count..

Nidhi Kushwaha,

I think you should count the number of time a particular person listen an audio, for how log time he/she listens and what is the trend of other listeners for that audio. Pandora is an example of song recommender system, you should study that as well. it would help you up to some extent.

Regards

Zafar Ali