cMTModel1.cpp

   1 // Implementation of the cMTModel1 class
   2 // Copyright © 2009 The University of Chicago
   3 #include "cMTModel1.h"
   4
   5 #include <QMessageBox>
   6 #include "mTVolca.h"
   7 #include "cMT.h"
   8 #include "Typedefs.h"
   9
  10 //////////////////////////////////////////////////////////////////////
  11 // Construction/Destruction
  12 //////////////////////////////////////////////////////////////////////
  13
  14 cMTModel1::cMTModel1(cMT* myMT, int Iterations)
  15         : m_myMT(myMT), m_Iterations(Iterations),
  16         m_T(),
  17         m_softCountOfT(),
  18         m_lamdaT() { }
  19
  20 cMTModel1::~cMTModel1() { }
  21
  22 void                    cMTModel1::initT()
  23 {
  24         mTVolca*                                                                myVolca;
  25         int                                                                             totalAssociatedLanguage2Words;
  26         IntToIntToDouble::iterator                              IntToIntToDoubleIt;
  27         IntToDouble::iterator                                   IntToDoubleIt;
  28         IntToDouble*                                                    oneListForOneLanguage1Word;
  29         double                                                                  uniformProb;
  30         int                                                                             key;
  31
  32
  33
  34         myVolca = m_myMT ->m_Volca;
  35
  36         m_T = myVolca ->m_fastWordsPairs;
  37         m_softCountOfT = myVolca ->m_fastWordsSoftCounts;
  38
  39
  40         for ( IntToIntToDoubleIt = m_T.begin(); IntToIntToDoubleIt != m_T.end();IntToIntToDoubleIt++)
  41         {
  42                 oneListForOneLanguage1Word = IntToIntToDoubleIt.data();
  43
  44                 totalAssociatedLanguage2Words = oneListForOneLanguage1Word ->size();
  45
  46                 uniformProb = 1.0 / totalAssociatedLanguage2Words;
  47
  48                 for ( IntToDoubleIt = oneListForOneLanguage1Word ->begin(); IntToDoubleIt != oneListForOneLanguage1Word ->end();  IntToDoubleIt++)
  49                 {
  50                         key = IntToDoubleIt.key();
  51                         (*oneListForOneLanguage1Word)[key] = uniformProb;
  52                 }
  53         }
  54
  55         for ( IntToIntToDoubleIt = m_softCountOfT.begin(); IntToIntToDoubleIt != m_softCountOfT.end();IntToIntToDoubleIt++)
  56         {
  57                 oneListForOneLanguage1Word = IntToIntToDoubleIt.data();
  58
  59                 for ( IntToDoubleIt = oneListForOneLanguage1Word ->begin(); IntToDoubleIt != oneListForOneLanguage1Word ->end();  IntToDoubleIt++)
  60                 {
  61                         IntToDoubleIt.data() = 0.0;
  62                 }
  63         }
  64
  65
  66         QMessageBox::information ( NULL, "Linguistica : MT Model1", "Finished InitT", "OK" );
  67
  68 }
  69
  70
  71 void                    cMTModel1::EMLoops(int numberOfIterations)
  72 {
  73         int                                                     loopI;
  74
  75         for (loopI =0; loopI < numberOfIterations; loopI++)
  76         {
  77                 // E step
  78                 EStep();
  79
  80                 //QMessageBox::information ( NULL, "Linguistica : MT Model1", "Finished E-Step", "OK" );
  81
  82                 // M step
  83                 MStep();
  84
  85                 //QMessageBox::information ( NULL, "Linguistica : MT Model1", "Finished M-Step", "OK" );
  86
  87                 // Clear softcouts for T
  88                 clearSoftCounts();
  89         }
  90
  91         // Release softcountT memory
  92         //releaseSoftCounts();
  93 }
  94
  95 void                    cMTModel1::clearSoftCounts()
  96 {
  97         IntToIntToDouble::iterator                              IntToIntToDoubleIt;
  98         IntToDouble*                                                    oneList;
  99         IntToDouble::iterator                                   IntToDoubleIt;
 100         int                                                                             key;
 101
 102         for ( IntToIntToDoubleIt = m_softCountOfT.begin(); IntToIntToDoubleIt != m_softCountOfT.end(); IntToIntToDoubleIt++)
 103         {
 104                 oneList = IntToIntToDoubleIt.data();
 105
 106                 for (IntToDoubleIt = oneList ->begin(); IntToDoubleIt != oneList ->end(); IntToDoubleIt++)
 107                 {
 108                         key = IntToDoubleIt.key();
 109                         (*oneList)[key] = 0.0;
 110                 }
 111         }
 112
 113 }
 114
 115 void                    cMTModel1::EStep()
 116 {
 117         mTVolca*                                                myVolca;
 118         int                                                             i;
 119
 120         myVolca = m_myMT ->m_Volca;
 121
 122         m_lamdaT.clear();
 123         for ( i=0; i < myVolca ->m_countOfSentences; i++)
 124         {
 125                 addSoftCountOfT(i);
 126         }
 127 }
 128
 129
 130 void                    cMTModel1::MStep()
 131 {
 132         IntToIntToDouble::iterator                              IntToIntToDoubleIt;
 133         IntToDouble*                                                    oneList;
 134         IntToDouble::iterator                                   IntToDoubleIt;
 135         int                                                                             language1Id;
 136         int                                                                             language2Id;
 137         double                                                                  oneTotalSoftCount;
 138         double                                                                  oneLamda;
 139
 140
 141         for ( IntToIntToDoubleIt = m_softCountOfT.begin(); IntToIntToDoubleIt != m_softCountOfT.end(); IntToIntToDoubleIt++)
 142         {
 143                 language1Id = IntToIntToDoubleIt.key();
 144                 oneLamda = m_lamdaT[language1Id] ;
 145                 oneList = IntToIntToDoubleIt.data();
 146
 147
 148                 for (IntToDoubleIt = oneList ->begin(); IntToDoubleIt != oneList ->end(); IntToDoubleIt++)
 149                 {
 150                         language2Id = IntToDoubleIt.key();
 151                         oneTotalSoftCount = IntToDoubleIt.data();
 152                         (*(m_T[language1Id]))[language2Id] = oneTotalSoftCount / oneLamda ;
 153                 }
 154         }
 155
 156
 157 }
 158
 159 void                    cMTModel1::addSoftCountOfT(int sentenceId)
 160 {
 161         mTVolca*                                                myVolca;
 162         double                                                  oneT;
 163         double                                                  deNumerator;
 164         int                                                             l,m;
 165         int                                                             language1WordId;
 166         int                                                             language2WordId;
 167         IntToInt*                                               oneLan1Sentence;
 168         IntToInt*                                               oneLan2Sentence;
 169
 170
 171         myVolca = m_myMT ->m_Volca;
 172
 173         oneLan1Sentence = myVolca ->m_language1Sentences[sentenceId];
 174         oneLan2Sentence = myVolca ->m_language2Sentences[sentenceId];
 175
 176
 177         for ( m=0; m< static_cast <int> (oneLan2Sentence ->size()); m++) //type cast to fix unsigned-signed warning
 178         {
 179                 language2WordId = (*oneLan2Sentence)[m];
 180
 181                 deNumerator =0;
 182
 183                 for ( l=0; l < static_cast<int> (oneLan1Sentence ->size()); l++) //type cast: unsigned-signed int comparison
 184                 {
 185                         language1WordId = (*oneLan1Sentence)[l];
 186
 187                         oneT = (*(m_T[language1WordId]))[language2WordId];
 188                         deNumerator += oneT;
 189                 }
 190
 191                 for ( l=0; l < static_cast <int> (oneLan1Sentence ->size()); l++)
 192                 {
 193
 194                         language1WordId = (*oneLan1Sentence)[l];
 195
 196                         oneT = (*(m_T[language1WordId]))[language2WordId] / deNumerator;
 197
 198                         (*(m_softCountOfT[language1WordId]))[language2WordId] += oneT;
 199
 200                         if ( m_lamdaT.contains(language1WordId))
 201                         {
 202                                 m_lamdaT[language1WordId] += oneT;
 203                         }
 204                         else
 205                         {
 206                                 m_lamdaT.insert(language1WordId, oneT);
 207                         }
 208                 }
 209
 210         }
 211
 212 }
 213
 214 void                    cMTModel1::releaseSoftCounts()
 215 {
 216         IntToIntToDouble::iterator                              IntToIntToDoubleIt;
 217         IntToDouble*                                                    oneList;
 218
 219         for ( IntToIntToDoubleIt = m_softCountOfT.begin(); IntToIntToDoubleIt != m_softCountOfT.end(); IntToIntToDoubleIt++)
 220         {
 221                 oneList = IntToIntToDoubleIt.data();
 222
 223                 delete oneList;
 224         }
 225
 226
 227 }