es.uam.eps.ir.ranksys.novdiv.reranking

Class LambdaReranker.LambdaUserReranker

java.lang.Object

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

es.uam.eps.ir.ranksys.novdiv.reranking.LambdaReranker.LambdaUserReranker

Enclosing class:

LambdaReranker<U,I>

protected abstract class LambdaReranker.LambdaUserReranker
extends GreedyReranker.GreedyUserReranker<U,I>

User re-ranker for LambdaReranker.

Field Summary

Fields

Modifier and Type	Field and Description
protected it.unimi.dsi.fastutil.objects.Object2DoubleMap<I>	novMap Map of the novelty of each item.
protected Stats	novStats Statistics about novelty scores.
protected Stats	relStats Statistics about relevance scores.

Fields inherited from class es.uam.eps.ir.ranksys.novdiv.reranking.GreedyReranker.GreedyUserReranker

maxLength, recommendation

Constructor Summary

Constructors

Constructor and Description
LambdaUserReranker(Recommendation<U,I> recommendation, int maxLength) Constructor.

Method Summary

Modifier and Type	Method and Description
protected double	norm(double score, Stats stats) Returns the normalized value of a relevance or novelty score.
protected abstract double	nov(Tuple2od<I> iv) Returns the novelty score of an item.
protected int	selectItem(it.unimi.dsi.fastutil.ints.IntSortedSet remainingI, List<Tuple2od<I>> list) Selects the next element of the permutation that maximizes the objective function.
protected double	value(Tuple2od<I> iv) Objective function that drives the greedy selection.

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

rerankPermutation, update

Methods inherited from class java.lang.Object

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

Field Detail

relStats

protected Stats relStats

Statistics about relevance scores.

novStats

protected Stats novStats

Statistics about novelty scores.

novMap

protected it.unimi.dsi.fastutil.objects.Object2DoubleMap<I> novMap

Map of the novelty of each item.

Constructor Detail

LambdaUserReranker

public LambdaUserReranker(Recommendation<U,I> recommendation,
                          int maxLength)

Constructor.

Parameters:

recommendation - input recommendation

maxLength - maximum length of the re-ranked recommendation

Method Detail

norm

protected double norm(double score,
                      Stats stats)

Returns the normalized value of a relevance or novelty score.

Parameters:

score - the relevance or novelty score

stats - the relevance or novelty statistics

Returns:

the normalized score

selectItem

protected int selectItem(it.unimi.dsi.fastutil.ints.IntSortedSet remainingI,
                         List<Tuple2od<I>> list)

Description copied from class: GreedyReranker.GreedyUserReranker

Selects the next element of the permutation that maximizes the objective function.

Overrides:

selectItem in class GreedyReranker.GreedyUserReranker<U,I>

Parameters:

remainingI - positions of the original recommendation that have not been selected yet.

list - the list of item-score pairs of the input recommendation

Returns:

the next element of the permutation that maximizes the objective function.

value

protected double value(Tuple2od<I> iv)

Description copied from class: GreedyReranker.GreedyUserReranker

Objective function that drives the greedy selection.

Specified by:

value in class GreedyReranker.GreedyUserReranker<U,I>

Parameters:

iv - item-score pair of the input recommendation.

Returns:

value of the function with the given item-score pair.

nov

protected abstract double nov(Tuple2od<I> iv)

Returns the novelty score of an item.

Parameters:

iv - item-relevance pair

Returns:

the novelty of the item