es.uam.eps.ir.ranksys.novelty.inverted.br

Class AlphaBayesRuleReranker<U,I>

java.lang.Object

es.uam.eps.ir.ranksys.novdiv.reranking.PermutationReranker<U,I>

es.uam.eps.ir.ranksys.novelty.inverted.br.BayesRuleReranker<U,I>

es.uam.eps.ir.ranksys.novelty.inverted.br.AlphaBayesRuleReranker<U,I>

Type Parameters:

U - type of the users

I - type of the items

All Implemented Interfaces:

Reranker<U,I>

public class AlphaBayesRuleReranker<U,I>
extends BayesRuleReranker<U,I>

Prior smoothed Bayesian probabilistic reformulation re-ranker. Estimation of likelihood and prior are based on the relevance scores of the recommender. S. Vargas and P. Castells. Improving sales diversity by recommending users to items.

Author:

Saúl Vargas (saul.vargas@uam.es)

Constructor Summary

Constructors

Constructor and Description
AlphaBayesRuleReranker(double alpha, it.unimi.dsi.fastutil.objects.Object2DoubleMap<I> norm) Constructor that receives a pre-calculated norm map.
AlphaBayesRuleReranker(double alpha, Stream<Recommendation<U,I>> recommendations) Constructor that receives a set of recommendations and calculates the norm.
AlphaBayesRuleReranker(double alpha, Stream<U> users, Recommender<U,I> recommender) Constructor that generates recommendations to calculate a norm.

Method Summary

Modifier and Type	Method and Description
static <U,I> it.unimi.dsi.fastutil.objects.Object2DoubleMap<I>	calculateNorm(Stream<Recommendation<U,I>> recommendations) Calculates the norm (sum of relevance scores) of each item.
protected double	likelihood(Tuple2od<I> iv) Returns the likelihood of an item: p(u | i).
protected double	prior(I i) Returns the prior of an item: p(i).

Methods inherited from class es.uam.eps.ir.ranksys.novelty.inverted.br.BayesRuleReranker

rerankPermutation

Methods inherited from class es.uam.eps.ir.ranksys.novdiv.reranking.PermutationReranker

getBasePerm, permuteRecommendation, rerankRecommendation

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

AlphaBayesRuleReranker

public AlphaBayesRuleReranker(double alpha,
                              it.unimi.dsi.fastutil.objects.Object2DoubleMap<I> norm)

Constructor that receives a pre-calculated norm map.

Parameters:

alpha - smoothing of the prior

norm - norm map

AlphaBayesRuleReranker

public AlphaBayesRuleReranker(double alpha,
                              Stream<Recommendation<U,I>> recommendations)

Constructor that receives a set of recommendations and calculates the norm.

Parameters:

alpha - smoothing of the prior

recommendations - previously calculated recommendations

AlphaBayesRuleReranker

public AlphaBayesRuleReranker(double alpha,
                              Stream<U> users,
                              Recommender<U,I> recommender)

Constructor that generates recommendations to calculate a norm.

Parameters:

alpha - smoothing of the prior

users - set of users to generate recommendations to calculate the norm

recommender - recommender whose norm is used

Method Detail

calculateNorm

public static <U,I> it.unimi.dsi.fastutil.objects.Object2DoubleMap<I> calculateNorm(Stream<Recommendation<U,I>> recommendations)

Calculates the norm (sum of relevance scores) of each item.

Type Parameters:

U - type of the users

I - type of the items

Parameters:

recommendations - set of recommendations used to calculate the norm

Returns:

item-norm map

likelihood

protected double likelihood(Tuple2od<I> iv)

Description copied from class: BayesRuleReranker

Returns the likelihood of an item: p(u | i).

Specified by:

likelihood in class BayesRuleReranker<U,I>

Parameters:

iv - item-relevance pair

Returns:

likelihood

prior

protected double prior(I i)

Description copied from class: BayesRuleReranker

Returns the prior of an item: p(i).

Specified by:

prior in class BayesRuleReranker<U,I>

Parameters:

i - item

Returns:

prior