Hi, brilliant work with the recommender modules. I've just started reading through the documentation for the API as well as Apache Mahout, but I haven't figured out how to determine which content has been used to compute a given positive recommendation. For example, I would like to display to the user some text on a page they rated highly, like "You may also like these other nodes, since two other people [with similar affinities] also rated them highly". How would that work?

Comments

danithaca’s picture

danithaca commented
Component: Miscellaneous » Code

This has to go through Mahout "explain" process. If this is a desirable feature for many people, I can add it to the Recommender API later.

dahacouk’s picture

dahacouk commented

This sounds like a great idea. We have a big requirement for this Mahout "explain" process. So, a big +1.

Can you point me to any other resources/documentation regarding the Mahout "explain" process? As I'd really like to read up on it and haven't been able to find anything at a first glance.

danithaca’s picture

danithaca commented

Just read the Mahout "explain" documentation for more info. They also have a friendly user mailing list.

danithaca’s picture

danithaca commented
Title: How to obtain which nodes were used to compute the similarity » Use "recommendBecause()" to obtain which nodes were used to compute the similarity

This requires a Mahout patch to UserBasedRecommender.recommendBecause()

danithaca’s picture

danithaca commented
Issue summary: View changes

Release 7.x-3.x is not supported anymore. Migrated the issue to 7.x-6.x.

danithaca’s picture

danithaca commented
Version: 7.x-3.x-dev » 7.x-6.0-alpha1