BiphoneCollection.cpp

   1 // Implementation of CBiphoneCollection methods
   2 // Copyright © 2009 The University of Chicago
   3 #include "BiphoneCollection.h"
   4 #include <cmath>
   5 #include "linguisticamainwindow.h"
   6 #include "ui/Status.h"
   7 #include "Lexicon.h"
   8 #include "Biphone.h"
   9 #include "Phone.h"
  10 #include "WordCollection.h"
  11 #include "log2.h"
  12
  13 // construction/destruction.
  14
  15 CBiphoneCollection::CBiphoneCollection(CWordCollection* words)
  16         : Q3Dict<CBiphone>(571),
  17         m_PtrArray(),
  18         m_CountSortArray(),
  19         m_MISortArray(),
  20         m_WMISortArray(),
  21         m_SortStyle(WMI),
  22         m_Phones(),
  23         m_MyWords(words),
  24         m_Count(),
  25         m_LogFreqs(),
  26         m_Freqs(),
  27         m_Length(0),
  28         m_MI(),
  29         m_SumOfMyMIs(0.0),
  30         m_TotalCount(0),
  31         m_Z_biphones(0.0)
  32 {
  33         // XXX. set auto-delete?
  34 }
  35
  36 CBiphoneCollection::CBiphoneCollection(CPhoneCollection* phones)
  37         : Q3Dict<CBiphone>(571),
  38         m_PtrArray(),
  39         m_CountSortArray(),
  40         m_MISortArray(),
  41         m_WMISortArray(),
  42         m_SortStyle(MI),
  43         m_Phones(phones),
  44         m_MyWords(),
  45         m_Count(),
  46         m_LogFreqs(),
  47         m_Freqs(),
  48         m_Length(0),
  49         m_MI(),
  50         m_SumOfMyMIs(0.0),
  51         m_TotalCount(0),
  52         m_Z_biphones(0.0)
  53 {
  54         // XXX. set auto-delete?
  55 }
  56
  57 CBiphoneCollection::~CBiphoneCollection()
  58 {
  59         delete[] m_MI;
  60         delete[] m_Freqs;
  61         delete[] m_LogFreqs;
  62         delete[] m_Count;
  63 }
  64
  65 void CBiphoneCollection::Empty()
  66 {
  67         clear();
  68         m_TotalCount = 0;
  69         m_Z_biphones = 0;
  70 }
  71
  72 CBiphone* CBiphoneCollection::Insert (CPhone* P1, CPhone* P2)
  73 {
  74
  75         CBiphone*               pBiphone;
  76         QString                 biphone;
  77         biphone = P1->Display() + "." + P2->Display();
  78
  79         m_TotalCount++;
  80
  81         pBiphone = find (biphone);
  82
  83         if ( ! pBiphone )
  84         {
  85                 pBiphone = new CBiphone(biphone);
  86                 pBiphone->m_MyFirst = P1;
  87                 pBiphone->m_MySecond = P2;
  88                 pBiphone->IncrementCorpusCount(1);
  89                 insert(biphone, pBiphone);
  90         }
  91         else
  92         {
  93                 pBiphone->IncrementCorpusCount(1);
  94         }
  95         return pBiphone;
  96
  97 }
  98 CBiphone* CBiphoneCollection::operator ^= (QString biphone)
  99 {
 100         return find (biphone);
 101 }
 102
 103 CBiphone* CBiphoneCollection::GetBiphone (CPhone* P1, CPhone* P2)
 104 {
 105
 106
 107         QString                 biphone;
 108         biphone = P1->Display() + "." + P2->Display();
 109
 110         return find (biphone);
 111
 112 }
 113 /*
 114 CBiphone* CBiphoneCollection::operator^= (CParse& Biphone)
 115 {
 116
 117         StringToBiphone::Iterator it = m_Hash.find(Biphone.Display());
 118
 119         if ( it == m_Hash.end() )
 120         {
 121                 return NULL;
 122         }
 123         else return &it.data();
 124
 125
 126   }
 127
 128 CBiphone*       CBiphoneCollection::GetBiphone (CPhone* Ph1, CPhone* Ph2)
 129 {
 130         CBiphone* pBiphone;
 131         QString Biphone = Ph1->Display() + "." + Ph2->Display();
 132
 133         StringToBiphone::Iterator it = m_Hash.find( Biphone );
 134
 135         if ( it == m_Hash.end() )
 136         {
 137                 return NULL;
 138         }
 139         else return pBiphone;
 140
 141
 142 }
 143 */
 144 void CBiphoneCollection::SetSize(int n)
 145 {
 146         delete[] m_Count;
 147         delete[] m_MI;
 148         delete[] m_Freqs;
 149         delete[] m_LogFreqs;
 150         m_LogFreqs = new double [n*n];
 151         m_Freqs = new double [n*n];
 152         m_MI = new double [n*n];
 153         m_Count = new int [n*n];
 154
 155         for (int i = 0; i < n*n; ++i) {
 156                 m_LogFreqs[i] = 0;
 157                 m_Freqs[i] = 0;
 158                 m_MI[i] = 0;
 159                 m_Count[i] = 0;
 160         }
 161         m_Length = n;
 162 }
 163
 164 void CBiphoneCollection::SetAtLogFreq(int FirstIndex, int SecondIndex, double ThisLogFreq)
 165 {
 166         m_LogFreqs[ FirstIndex * m_Length + SecondIndex ] = ThisLogFreq;
 167 }
 168 double CBiphoneCollection::GetLogFreq(int FirstIndex, int SecondIndex)
 169 {
 170         return m_LogFreqs[ FirstIndex * m_Length + SecondIndex ];
 171 }
 172 void CBiphoneCollection::SetAtCount(int FirstIndex, int SecondIndex, int ThisCount)
 173 {
 174         m_Count[ FirstIndex * m_Length + SecondIndex ] = ThisCount;
 175 }
 176 int CBiphoneCollection::GetCount(int FirstIndex, int SecondIndex)
 177 {
 178         return m_Count[ FirstIndex * m_Length + SecondIndex ];
 179 }
 180
 181 void CBiphoneCollection::SetAtMI(int FirstIndex, int SecondIndex, double ThisMI)
 182 {
 183         m_MI[ FirstIndex * m_Length + SecondIndex ] = ThisMI;
 184 }
 185
 186
 187 void CBiphoneCollection::IncrementAtCount (int FirstPhoneIndex, int SecondPhoneIndex, int n)
 188 {
 189         m_Count[ FirstPhoneIndex * m_Length + SecondPhoneIndex ] += n;
 190 }
 191
 192 void CBiphoneCollection::ListDisplay( Q3ListView* pView )
 193 {
 194         pView->setRootIsDecorated( FALSE );
 195         pView->setSorting(1);
 196         // Remove all previous columns
 197         while( pView->columns() ) pView->removeColumn( 0 );
 198
 199         //    Add Column headers
 200         pView->addColumn( "Bigram" );
 201         pView->addColumn( "Count" );
 202         pView->addColumn( "Frequency" );
 203         pView->addColumn ( "Log Cond Prob" );
 204         pView->addColumn( "+LogProb" );
 205         pView->addColumn( "MutualInf" );
 206         pView->addColumn( "WMI" );
 207         pView->addColumn( "Cond Prob" );
 208         pView->addColumn( "MI markedness" );
 209
 210         CLexicon& lex = *m_MyWords->GetLexicon();
 211         linguistica::ui::status_user_agent& status = lex.status_display();
 212
 213         Q3Dict<CBiphone>& dict = *this;
 214         status.major_operation = "Creating biphone list for display";
 215         status.progress.clear();
 216         status.progress.set_denominator(dict.count());
 217
 218         int biphone_nr = 0;
 219         Q3DictIterator<CBiphone> iter(dict);
 220         while (CBiphone* biphone = iter()) {
 221                 biphone->BiphoneListDisplay(pView);
 222                 status.progress = biphone_nr++;
 223         }
 224
 225         status.progress.clear();
 226         status.major_operation.clear();
 227 }
 228
 229 double CBiphoneCollection::GetSumOfMyMIs()
 230 {
 231         Q3DictIterator<CBiphone> it (*this);
 232         CBiphone* pBiphone;
 233         if ( m_SumOfMyMIs == 0 )
 234         {
 235                 for ( ; it.current(); ++it)
 236                 {
 237                         pBiphone = it.current();
 238                         m_SumOfMyMIs += pBiphone->m_MI;
 239                 }
 240         }
 241         return m_SumOfMyMIs;
 242 }
 243 double CBiphoneCollection::ComputeZ_MI()
 244 {
 245         using std::pow;
 246
 247         //      Currently this is not correctly computing this Z:
 248         //  To compute it correctly, we need to assign an MI value to the unseen bigrams
 249         //  and take those into consideration. By leaving them out, we're assuming
 250         //  an infinite negative MI for those bigrams.
 251         m_Z_biphones = 0;
 252         Q3DictIterator<CBiphone> it (*this);
 253         CBiphone* pBiphone;
 254         double  sum =  0;
 255         if ( m_Z_biphones == 0 )
 256         {
 257                 for ( ; it.current(); ++it)
 258                 {
 259                         pBiphone = it.current();
 260                         sum += pow(2, -1 * pBiphone->m_MI);
 261                 }
 262         }
 263         m_Z_biphones = log2(sum);
 264         return m_Z_biphones;
 265 }