Stem_EncodingLength.cpp

   1 // Implementation of CStem description length methods
   2 // Copyright © 2009 The University of Chicago
   3 #include "Stem.h"
   4 #include "MiniLexicon.h"
   5 #include "Lexicon.h"
   6 #include "StemCollection.h"
   7 #include "log2.h"
   8
   9 double CStem::GetLengthOfPointerToMe() const
  10 {
  11         if (m_LengthOfPointerToMe > 0)
  12                 return m_LengthOfPointerToMe;
  13         m_LengthOfPointerToMe = log2(m_pMyMini->GetStems()->GetCorpusCount() /
  14                                      GetCorpusCount());
  15         return m_LengthOfPointerToMe;
  16 }
  17
  18 double CStem::GetLengthOfPointerToMe_2()
  19 { return GetLengthOfPointerToMe(); }
  20
  21 double CStem::GetPhonologicalInformationContent(CLexicon* MotherLexicon) const
  22 {
  23         if (m_PhonologicalContent != 0.0)
  24                 return m_PhonologicalContent;
  25
  26         if (MotherLexicon != 0)
  27                 CalculatePhonologicalInformationContent(MotherLexicon);
  28         return m_PhonologicalContent;
  29 }
  30
  31 double CStem::CalculatePhonologicalInformationContent(CLexicon* Lexicon) const
  32 {
  33         m_PhonologicalContent = m_BigramLogProb > 0.0 ?
  34                 m_BigramLogProb :
  35                 CParse::ComputeDL(Lexicon->GetNumberOfCharacterTypes());
  36         return m_PhonologicalContent;
  37 }
  38
  39 // Calculate the description length
  40 //
  41 // Returns:
  42 //    float - the descrition length
  43 float CStem::CalculateDL() const
  44 {
  45         static const float PointerLength = log2(float(27.0));
  46         return PointerLength * GetKeyLength();
  47 }