pferrel
commented
8 years ago Please post questions to https://groups.google.com/forum/#!forum/actionml-user https://groups.google.com/forum/#!forum/actionml-user
Context is just another indicator, (user-id, event-name, item-id) anything that can be encoded in this way is a candidate for use. The event-name and item-id correspond so (user-id, “running”, music-id) might be a candidate indicator or even (user-id, “status”, “running”). The strongest way to encode this would need to be tested with out analysis tools but the way the UR works, a bad indicator-type will do no harm.
On Aug 28, 2016, at 2:58 AM, Subhash Ramesh notifications@github.com wrote:
Hi,
How does the UR template allow to recommend based on user context? What query and events allow for this?
For example, let's say I'm making a music app. In this case, my primary indicator would be "play60" (play 60% of the music, for instance), and there would be many secondary indicators (like, dislike, ban, follow, etc.)
When a user plays a track, the event that I record contains many contextual data attached. For instance, a play event might contain the time (morning, afternoon, evening) the user played the track, the type of music, the user's activity (running, walking, driving), the user's location, weather, and other music's attributes.
So, how do I create these events with contextual data? Will the UR automatically learn that a specific user tends to listen to relaxing music in the evening for instance? How does it accept contextual data?
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/pferrel/template-scala-parallel-universal-recommendation/issues/18, or mute the thread https://github.com/notifications/unsubscribe-auth/AAT8S8h6tNmx0wZ0N-559iwDMbEZFG0Dks5qkVvKgaJpZM4Ju46K.
Hi,
How does the UR template allow to recommend based on user context? What query and events allow for this?
For example, let's say I'm making a music app. In this case, my primary indicator would be "play60" (play 60% of the music, for instance), and there would be many secondary indicators (like, dislike, ban, follow, etc.)
When a user plays a track, the event that I record contains many contextual data attached. For instance, a play event might contain the time (morning, afternoon, evening) the user played the track, the type of music, the user's activity (running, walking, driving), the user's location, weather, and other music's attributes.
So, how do I create these events with contextual data? Will the UR automatically learn that a specific user tends to listen to relaxing music in the evening for instance? How does it accept contextual data?