Main Page   Namespace List   Class Hierarchy   Alphabetical List   Compound List   File List   Namespace Members   Compound Members   File Members   Related Pages  

SimpleKLDocModel Class Reference

Doc representation for simple KL divergence retrieval model. More...

#include <SimpleKLDocModel.hpp>

Inheritance diagram for SimpleKLDocModel:

DocumentRep AbsoluteDiscountDocModel DirichletPriorDocModel JelinekMercerDocModel TwoStageDocModel List of all members.

Public Methods

 SimpleKLDocModel (DOCID_T docID, const UnigramLM &collectLM, int dl=1, const double *prMass=NULL, SimpleKLParameter::SmoothStrategy strat=SimpleKLParameter::INTERPOLATE)
 ~SimpleKLDocModel ()
virtual double termWeight (TERMID_T termID, const DocInfo *info) const
 term weighting function, weight(w) = p_seen(w)/p_unseen(w)

virtual double scoreConstant () const
 doc-specific constant term in the scoring formula

virtual double unseenCoeff () const=0
 a(d)

virtual double seenProb (double termFreq, TERMID_T termID) const=0
 p(w|d), w seen


Protected Attributes

const UnigramLMrefLM
const double * docPrMass
SimpleKLParameter::SmoothStrategy strategy

Detailed Description

Doc representation for simple KL divergence retrieval model.

abstract interface of doc representation for smoothed document unigram model

adapt a smoothed document language model interface to a DocumentRep interface

 p(w|d) = q(w|d) if w seen
        = a(d) * Pc(w)  if w unseen
   where,  a(d) controls the probability mass allocated to all unseen words and     Pc(w) is the collection language model


Constructor & Destructor Documentation

SimpleKLDocModel::SimpleKLDocModel DOCID_T    docID,
const UnigramLM   collectLM,
int    dl = 1,
const double *    prMass = NULL,
SimpleKLParameter::SmoothStrategy    strat = SimpleKLParameter::INTERPOLATE
[inline]
 

SimpleKLDocModel::~SimpleKLDocModel   [inline]
 


Member Function Documentation

virtual double SimpleKLDocModel::scoreConstant   const [inline, virtual]
 

doc-specific constant term in the scoring formula

Implements DocumentRep.

virtual double SimpleKLDocModel::seenProb double    termFreq,
TERMID_T    termID
const [pure virtual]
 

p(w|d), w seen

Implemented in JelinekMercerDocModel, DirichletPriorDocModel, AbsoluteDiscountDocModel, and TwoStageDocModel.

virtual double SimpleKLDocModel::termWeight TERMID_T    termID,
const DocInfo   info
const [inline, virtual]
 

term weighting function, weight(w) = p_seen(w)/p_unseen(w)

Implements DocumentRep.

virtual double SimpleKLDocModel::unseenCoeff   [pure virtual]
 

a(d)

Implemented in JelinekMercerDocModel, DirichletPriorDocModel, AbsoluteDiscountDocModel, and TwoStageDocModel.


Member Data Documentation

const double* SimpleKLDocModel::docPrMass [protected]
 

const UnigramLM& SimpleKLDocModel::refLM [protected]
 

SimpleKLParameter::SmoothStrategy SimpleKLDocModel::strategy [protected]
 


The documentation for this class was generated from the following file:
Generated on Wed Nov 3 12:59:55 2004 for Lemur Toolkit by doxygen1.2.18