search

PredictionIO / template-scala-parallel-universal-recommendation

PredictiionIO Template for Universal Recommender
111 stars 48 forks source link

Multiple Item types #49

Open techscientist opened 8 years ago

techscientist commented 8 years ago

Hi,

Does this template support multiple item types? For example, let's say this is a music app. Can we create events for users liking artists, playing tracks, playing albums, disliking tracks, etc.?

How would we do this?

pferrel commented 8 years ago

Public support in at https://groups.google.com/forum/#!forum/actionml-user https://groups.google.com/forum/#!forum/actionml-user

All this data can be used in the Universal Recommender and is why it was created. But you can recommend only one item type, and you chose that when you build a recommender model. With your data you can build several models with identical data each to recommend something different—some different item-type, for instance artists, tracks, albums. Dislikes can often help predict likes so all data can be used.

On Aug 27, 2016, at 3:07 AM, Subhash Ramesh notifications@github.com wrote:

Hi,

Does this template support multiple item types? For example, let's say this is a music app. Can we create events for users liking artists, playing tracks, playing albums, disliking tracks, etc.?

How would we do this?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/PredictionIO/template-scala-parallel-universal-recommendation/issues/49, or mute the thread https://github.com/notifications/unsubscribe-auth/AAT8S_QY_8lSXm9C9JFjFYvrMV2vlifVks5qkAxUgaJpZM4Juqdb.

techscientist commented 8 years ago

Oh ok, But, would just adding a "type" property to the event, and then query using bias of the type = -1 work, since then, only recommendations with the type being equal to the queried type (artist, playlist, track, etc.) will be returned?