Configuring the model

The first step is to choose a model/algorithm for the learning procedure and set it's parameters.

The training set

Once the model is configured. You can create a training set. The training set is obtained from the past users behavior or from items ratings.

Learning

Use the training set to capture patterns and to identify factors that are important for a user. In the same time measure the extent to which an item possesses some feature.

References

• http://en.wikipedia.org/wiki/Recommendation_system
• http://public.research.att.com/~volinsky/netflix/kdd08koren.pdf
• http://research.yahoo.com/files/ieeecomputer.pdf