next up previous
Next: Weighted Sum Up: Item-based Collaborative Filtering Algorithm Previous: Adjusted Cosine Similarity

Prediction Computation

The most important step in a collaborative filtering system is to generate the output interface in terms of prediction. Once we isolate the set of most similar items based on the similarity measures, the next step is to look into the target users' ratings and use a technique to obtain predictions. Here we consider two such techniques.


Badrul M. Sarwar