Package org.terrier.matching.models.normalisation

Provides the interface and the classes for implementing the basic models for randomness in the DFR framework.

See: Description

Class Summary

Class	Description
Normalisation	This class provides a contract for implementing frequency normalisation methods.
Normalisation0	This class implements an empty normalisation.
Normalisation1	This class implements the DFR normalisation 1, which is identical to DFR normalisation 2 with c=1.0
Normalisation2	This class implements the DFR normalisation 2.
Normalisation2exp	This class implements the DFR normalisation 2 with natural logorithm.
Normalisation3	This class implements the Dirichlet Priors normalisation.
NormalisationB	This class implements BM25's normalisation method.
NormalisationF	This class implements an increasing density function for the frequency normalisation.
NormalisationJ	This class implements the tf normalisation based on Jelinek-Mercer smoothing for language modelling.
NormalisationJN	This class implements the tf normalisation based on Jelinek-Mercer smoothing for language modelling where collection model is given by document frequency instead of term frequency.
NormalisationP	This class implements Term Frequency Normalisation via Pareto Distributions
NormalisationStatic	This class implements a Normalisation method that forces all term frequencies to the value of the parameter.

Package org.terrier.matching.models.normalisation Description

Provides the interface and the classes for implementing the basic models for randomness in the DFR framework. The implemented classes can be used by specifying the class name in /etc/trec.models, or in property trec.model. For instance, adding a line DFRWeightingModel(P,L,2) in file /etc/trec.models tells the systems to use the PL2 model to run retrieval. P stands for class P in this package, L stands for class L in package org.terrier.matching.basicmodel, and "2" stands for class 2 in package org.terrier.matching.models.normalisation.