PhoneCollection.cpp

   1 // Implementation of CPhoneCollection methods
   2 // Copyright © 2009 The University of Chicago
   3 #include "PhoneCollection.h"
   4
   5 #include <Q3TextStream>
   6 #include <QIODevice>
   7 #include <QFile>
   8 #include "ui/Status.h"
   9 #include "MonteCarlo.h"
  10 #include "Phone.h"
  11 #include "Stem.h"
  12 #include "BiphoneCollection.h"
  13 #include "WordCollection.h"
  14 #include "log2.h"
  15
  16 CPhoneCollection::CPhoneCollection(CWordCollection* words)
  17         : m_MyWords(words),
  18         m_MyBiphones(words),
  19         m_expMI(),
  20         m_expMIFromBoundary(),
  21         m_expMIToBoundary() { }
  22
  23 double CPhoneCollection::GetSumOfMyMIs()
  24         { return m_MyBiphones.GetSumOfMyMIs(); }
  25
  26 void CPhoneCollection::CountPhonesAndBiphonesInWord( CStem* pStem, eTier WhichTier)
  27 {
  28
  29         CPhone*                 pPhone, *prevPhone;
  30         prevPhone = NULL;
  31         CParse*                 pThisString;
  32         QString                 Biphone;
  33         switch (WhichTier)
  34         {
  35         case (TIER_1):
  36                 pThisString = pStem->GetPhonology_Tier1();
  37                 break;
  38         case (TIER_1_SKELETON):
  39                 pThisString = pStem->GetPhonology_Tier1_Skeleton();
  40                 break;
  41         default:
  42                 pThisString = pStem->GetPhonology_Tier2();
  43         }
  44
  45
  46
  47         if (pThisString->Size()==0) return;
  48
  49         for (int phoneno =1; phoneno <=  pThisString->Size(); phoneno++)
  50         {
  51                 QString temp2 = pThisString->GetPiece(phoneno).Display();
  52
  53                 if ( phoneno  == pThisString->Size() &&         // these conditions define a # at the end of a string like #dog#
  54                          pThisString->GetPiece(1).Display() == QString("#")                                     &&
  55                          pThisString->GetPiece(phoneno).Display() == QString("#")
  56                         )
  57                         {
  58                                 pPhone = *this ^= pThisString->GetPiece(phoneno);
  59                         }
  60                         else
  61                         {
  62                                 pPhone = *this << pThisString->GetPiece(phoneno);
  63                         }
  64                         if (phoneno > 1)
  65                                 m_MyBiphones.Insert(prevPhone, pPhone);
  66                         prevPhone = pPhone;
  67         }
  68 }
  69
  70 void CPhoneCollection::Normalize()
  71 {
  72         CPhone*                 pPhone;
  73         CBiphone*               pBiphone;
  74         double                  MI, Freq;
  75
  76         for (int phoneno = 0; phoneno < GetCount(); phoneno++)
  77         {
  78                 pPhone = GetAt(phoneno);
  79                 pPhone->m_Frequency = pPhone->GetCorpusCount() / (double) m_CorpusCount;
  80                 pPhone->m_LogFreq = -1 * base2log( pPhone->m_Frequency );
  81         }
  82
  83
  84         Q3DictIterator<CBiphone> it(m_MyBiphones);
  85         for ( ; it.current(); ++it)
  86         {
  87                 pBiphone = it.current();
  88                 Freq = (double) pBiphone->GetCorpusCount() / m_MyBiphones.GetTotalCount();
  89                 pBiphone->m_Freq = Freq;
  90                 pBiphone->m_LogFreq =  -1 * base2log ( Freq );
  91                 pBiphone->m_CondProb = Freq / (double) pBiphone->m_MyFirst->m_Frequency;
  92                 pBiphone->m_LogCondProb = -1 * base2log ( pBiphone->m_CondProb );
  93
  94                 MI = base2log ( Freq / ( pBiphone->m_MyFirst->m_Frequency * pBiphone->m_MySecond->m_Frequency ) );
  95
  96                 pBiphone->m_MI = MI;
  97                 pBiphone->m_WMI = MI * pBiphone->GetCorpusCount();
  98                 pBiphone->m_NormalizedMI = pBiphone->m_MI - m_MyBiphones.m_Z_biphones;
  99         }
 100         m_MyBiphones.ComputeZ_MI();
 101
 102         for (it.toFirst() ; it.current(); ++it)
 103         {
 104                 pBiphone = it.current();
 105                 pBiphone->m_NormalizedMI =
 106                         m_MyBiphones.m_Z_biphones - pBiphone->m_MI;
 107         }
 108 }
 109
 110 void CPhoneCollection::ListDisplay(Q3ListView* pView,
 111         linguistica::ui::status_user_agent& status)
 112 {
 113         CPhone* pPhone;
 114
 115         pView->setRootIsDecorated(false);
 116         pView->setSorting(1);
 117         // Remove all previous columns
 118         while (pView->columns() != 0)
 119                 pView->removeColumn(0);
 120
 121         // Add Column headers
 122         pView->addColumn("Phone");
 123         pView->addColumn("Count");
 124         pView->addColumn("+LogProb");
 125         pView->addColumn("Prob");
 126
 127         if (m_SortValidFlag == false)
 128                 Sort(KEY);
 129
 130         // Display all items
 131         status.major_operation = "Creating phone list for display";
 132         status.progress.clear();
 133         status.progress.set_denominator(GetCount());
 134         for (int phoneno = 0; phoneno < (int) GetCount(); phoneno++) {
 135                 pPhone = GetAtSort(phoneno);
 136                 pPhone->PhoneListDisplay(pView);
 137                 status.progress = phoneno;
 138         }
 139         status.progress.clear();
 140         status.major_operation.clear();
 141 }
 142
 143 void CPhoneCollection::PopulateMonteCarlo( MonteCarlo* pMyMonteCarlo)
 144 {
 145         QString                         FirstPhone;
 146         MonteCarlo*                     qMonteCarlo;
 147         int                                     Size;
 148         int                                     NumberOfBigramsFound = 0;
 149         CPhone*                         pPhone, *qPhone;
 150         CBiphone*                       pBiphone;
 151         int                                     j;
 152         // first, Unigram frequencies:
 153         for (int i = 0; i < GetSize(); i++)
 154         {
 155                 pPhone = GetAt(i);
 156                 pMyMonteCarlo->StockDictionary( pPhone->Display(), pPhone->m_Frequency, i );
 157         }
 158
 159         static const QString FileName = "c:\\LxaDeleteMe.txt";
 160         QFile file( FileName );
 161         Q3TextStream outf( &file );
 162         outf.setEncoding( Q3TextStream::Unicode );
 163
 164         // Now Bigram frequencies:
 165         if (pMyMonteCarlo->m_ModelType == BIGRAM)
 166         {
 167                 for (int i = 0; i < GetSize(); i++) // iterate throught the phonemes...
 168                 {
 169                         pPhone                  = GetAt(i);
 170                         FirstPhone              = pPhone->Display();
 171                         Size                    = GetSize();
 172                         qMonteCarlo             = new MonteCarlo ( Size, FirstPhone );
 173                         pMyMonteCarlo   ->GetMyBigrams()->insert ( FirstPhone, qMonteCarlo ); // this is a QDict within the MonteCarlo, for subMonteCarlos for each phone
 174                         NumberOfBigramsFound = 0;
 175
 176                         outf << endl << FirstPhone << endl;
 177
 178                         for (j = 0; j < Size; j++)  // iterate through the phonemes, for the second phoneme of this bigram
 179                         {
 180                                 qPhone          = GetAt(j);
 181                                 pBiphone        = m_MyBiphones.GetBiphone(pPhone, qPhone);
 182                                 if (pBiphone)
 183                                 {
 184                                         qMonteCarlo->StockDictionary( qPhone->Display(), pBiphone->m_CondProb, j);
 185                                         NumberOfBigramsFound++;
 186                                 }
 187                         }
 188                         qMonteCarlo->SetSize ( NumberOfBigramsFound );
 189                         qMonteCarlo->Normalize();
 190                         //qMonteCarlo->Dump( &outf );
 191
 192
 193
 194
 195
 196
 197                 }
 198         }
 199
 200         pMyMonteCarlo->Normalize();
 201
 202
 203
 204
 205
 206
 207         file.close();
 208
 209
 210 }
 211
 212 void CPhoneCollection::ComputeStringAgreementAndDisagreement(
 213         CLParse* string1, CLParse* string2,
 214         double& agreement_unigram, double& agreement_bigram,
 215         double& disagreement_unigram, double& disagreement_bigram)
 216 {
 217         struct not_implemented { };
 218         throw not_implemented();
 219
 220         static_cast<void>(string1);
 221         static_cast<void>(string2);
 222         static_cast<void>(agreement_unigram);
 223         static_cast<void>(agreement_bigram);
 224         static_cast<void>(disagreement_unigram);
 225         static_cast<void>(disagreement_bigram);
 226
 227 //      QString string1_alpha = string1->GetAlphabetizedForm();
 228 //      QString string2_alpha = string2->GetAlphabetizedForm();
 229 //      int i,j;
 230 //      j = 1;
 231 //      // first calculate unigram overlap and non-overlap costs.
 232 //      for (i=1; i <= string1->Size(); i++)
 233 //      {
 234 //              if (string1_alpha[i] == string2_alpha[j] )
 235 //              {
 236 //                      agreement_unigram += ;
 237 //              }
 238 //              else if (string1_alpha[i] < string2_alpha[j])
 239 //              {
 240 //                      disagreement_unigram += ;
 241 //                      if (i==string1->Size()){
 242 //                              break;
 243 //                      }
 244 //                      i++;
 245 //              }
 246 //              else
 247 //              {
 248 //                      disagreement_unigram += ;
 249 //                      if (j==string1->Size()){
 250 //                              break;
 251 //                      }
 252 //                      j++;
 253 //              }
 254 //      }
 255 //      if (i < string1->Size()){
 256 //              for (; i <= string1->Size(); i++){
 257 //                      disagreement_unigram +=  ;
 258 //              }
 259 //      }
 260 //      if (j < string1->Size()){
 261 //              for (; j <= string1->Size(); j++){
 262 //                      disagreement_unigram +=  ;
 263 //              }
 264 //      } // end of calculation of unigram figures.
 265 }