-
Book Overview & Buying
-
Table Of Contents
-
Feedback & Rating

AI Blueprints
By :

With content-based recommendations, as described in the preceding section, we only use the items' properties, such as their titles and descriptions, to generate recommendations of similar items. We demonstrated these kinds of recommendations with Amazon product data. The fact that users on Amazon are actually buying and reviewing the products makes no difference in content-based recommendations.
Collaborative filtering recommendations utilize only user activity. We can still find items similar to a specific item, but this time the similar items are found by finding items that are purchased or rated highly by users who also rated or purchased the item in question. Perhaps more importantly, we can also find recommendations for a particular user. Finding recommendations for a particular user is not possible with content-based recommendations. We can do this by finding similar users, based on rating or purchase history, and then determining which...