ivory.ffg.feature

Interface Feature

All Known Implementing Classes:

OrderedWindowSequentialDependenceFeature, TermFeature, UnorderedWindowSequentialDependenceFeature

public interface Feature

Abstract feature definition. Feature values are computed using a sliding window or an indexed document.

Author:

Nima Asadi

Method Summary

Methods

Modifier and Type	Method and Description
float	computeScoreWithMiniIndexes(int[][] positions, int[] query, int dl, GlobalStats stats) Computes the feature value using a (mini-)indexed document.
float	computeScoreWithSlidingWindow(int[] document, int[] query, int[] hashedQuery, GlobalStats stats) Computes the feature value using a sliding window given a document, query and a scoring function
void	initialize(ScoringFunction score) Initializes this feature.

Method Detail

computeScoreWithMiniIndexes

float computeScoreWithMiniIndexes(int[][] positions,
                                int[] query,
                                int dl,
                                GlobalStats stats)

Computes the feature value using a (mini-)indexed document.

Parameters:

positions - Positions of each term in the query

query - Original query terms

dl - Document length

stats - Global statistics

Returns:

Feature value

computeScoreWithSlidingWindow

float computeScoreWithSlidingWindow(int[] document,
                                  int[] query,
                                  int[] hashedQuery,
                                  GlobalStats stats)

Computes the feature value using a sliding window given a document, query and a scoring function

Parameters:

document - Flat array representation of a document vector.

query - Original query terms (used to retrieve global statistics)

hashedQuery - Hashed query terms (the same as query for techniques that don't use hashing)

stats - Global statistics

Returns:

Feature value

initialize

void initialize(ScoringFunction score)

Initializes this feature.

Parameters:

score - scoring function