Baseline Algorithms

LensKit provides several baseline algorithms based on simple computations such as item popularity and user/item biases.

The PopularityRankItemScorer class provides the basic item scoring logic for popularity-based recommendation. It is implemented as an item scorer so that popularity information can be incorporated into other scorers more easily.

To use it, bind it to the ItemScorer interface:

bind ItemScorer to PopularityRankItemScorer

The scores produced by this scorer are the percentile ranks for each item in the range [0,1], where the most popular item has a rank of 1 and unknown items have a rank of 0.

Bias Models

LensKit’s user/item bias models can be used to compute item scores:

bind ItemScorer to BiasItemScorer
bind BiasModel to LiveUserItemBiasModel

results matching ""

    No results matching ""