Recommendation cold start problem

User 1102 | 1/6/2015, 3:20:55 PM

how to recommend new user where there is no historical data or past behaviour is available ? how graph lab recommender works on it ?

Comments

User 19 | 1/6/2015, 7:08:39 PM

Hi,

The available recommender algorithms all leverage past behavior. However, if you have a small amount of it, you can augment your recommendations by using other information you have about your items and your users. Check out the documentation for the user_side_data and item_side_data arguments in <a href="http://graphlab.com/products/create/docs/generated/graphlab.recommender.create.html#graphlab.recommender.create">graphlab.recommender.create</a>. You can also read more about this in the <a href="http://graphlab.com/learn/userguide/index.html#ModelingdataRecommender_systems">recommender section of the user guide</a>.

If you have more questions, please let us know. Chris