SignatureCollection_PartsOfSpeech.cpp

   1 // Part of speech discovery
   2 // Copyright © 2009 The University of Chicago
   3 #include "MiniLexicon.h"
   4
   5 #include <memory>
   6 #include <QTextStream>
   7 #include <QIODevice>
   8 #include <QFile>
   9 #include <QString>
  10 #include "SignatureCollection.h"
  11 #include "Signature.h"
  12 #include "POS.h"
  13 #include "POSCollection.h"
  14 #include "Parse.h"
  15 #include "HTML.h"
  16
  17 QString HTMLfileStart("<html><head> "
  18         "<link rel=\"stylesheet\" type=\"text/css\" href=\"lxa.css\"> "
  19         "</head>"
  20         "<body><table>");
  21
  22 void CMiniLexicon::FindMajorSignatures()
  23 {
  24         // XXX. add more options, for prefixes, etc.
  25         // XXX. Make this a user-defined parameter
  26         const int NumberOfTopSignatures = 50;
  27         const int MinimumSignatureSize  = 2;
  28         const QString filename = "c:\\PartsOfSpeech.html";
  29
  30         // open log file
  31         QFile file(filename);
  32         if (!file.open(QIODevice::WriteOnly))
  33                 // open failed
  34                 return;
  35         QTextStream outf(&file);
  36
  37         if (m_pPOS == 0) m_pPOS = new LxPoSCollection;
  38
  39         outf << HTMLfileStart;
  40         outf << StartTableRow << TableData ("Signature") << EndTableRow;
  41
  42
  43         // For each signature: if it's good enough, make it a part of speech
  44         int TooBigToIgnoreRankCutoff = 10;
  45         m_pSignatures->FindDisplayOrdering();
  46         for (int signo = 0, count = 0;
  47                         signo < m_pSignatures->GetCount() &&
  48                         count < NumberOfTopSignatures;
  49                         ++signo) {
  50                 CSignature* pSig = m_pSignatures->GetAtSort(signo);
  51
  52                 outf << "<tr><td>" << pSig->Display('-') << "</td>";
  53
  54                 // hack: simulate labelled continue statement
  55                 struct not_eligible { };
  56                 try {
  57                         // a sig P is a "mentor" to sig Q if P is more robust
  58                         // than Q and P contains Q
  59                         if (pSig->GetMentor() != 0)
  60                                 // not a part of speech: subsumed under mentor
  61                                 throw not_eligible();
  62
  63                         if (pSig->Size() < MinimumSignatureSize)
  64                                 // not a part of speech: too simple
  65                                 throw not_eligible();
  66
  67                         if (pSig->GetMentorList()->count() < 2 && signo > TooBigToIgnoreRankCutoff )
  68                             // it has no mentees, and it is not among the very biggest signatures
  69                                 throw not_eligible();
  70
  71
  72                         for (LxPoS* qPOS = m_pPOS->first();
  73                                         qPOS != 0; qPOS=m_pPOS->next()) {
  74                                 // If pSig has exactly one more affix than a
  75                                 // signature already in PoS, then pSig can't
  76                                 // be a PoS, and its extra affix is entered as a satellite.
  77                                 CSignature* qSig = qPOS->GetSignature();
  78
  79                                 if (qSig->Size() + 1 == pSig->Size() &&
  80                                                 qSig->Contains (pSig)) {
  81                                         CParse Suffix = qSig->Intersection(*pSig);
  82                                         qSig->AppendSatelliteAffix (Suffix);
  83                                         outf<< "<td>" << Suffix.Display();
  84                                         throw not_eligible();
  85                                 }
  86                         }
  87                 } catch (not_eligible) { continue; }
  88
  89                 std::auto_ptr<LxPoS> new_pos(new LxPoS(pSig, this));
  90                 LxPoS* pPOS = new_pos.get();
  91
  92                 // Record affix frequencies
  93                 for (int k = 1; k <= pSig->Size(); ++k) {
  94                         pPOS->SetPieceValue( k, pSig->GetNumberOfStems() );
  95                         outf << "<td>"<<pSig->GetPiece(k).Display() << "</td>";
  96                 }
  97
  98                 // Calculate robustness
  99                 pPOS->AddRobustness(pSig->GetRobustness());
 100
 101                 // Save part-of-speech
 102                 m_pPOS->append(new_pos.release());
 103                 ++count;
 104
 105                 outf << "<td>"<< pSig->GetNumberOfStems()  <<"</td></tr>";
 106                 CSignature *qSig;
 107                 // For each mentee:
 108                 for (int signo = 0; signo < pSig->GetMentorList()->size(); signo++)
 109                               {       qSig = pSig->GetMentorList()->at(signo);
 110                                       outf << StartTableRow << TableData (qSig->Display('-')) ;
 111                                       pPOS->AppendSignature( qSig );
 112                                       // Adjust robustness for mentee.
 113                                       pPOS->AddRobustness( qSig->GetRobustness() );
 114
 115                                       //Adjust stem count
 116                                       for (int stemno = 0; stemno < qSig->GetNumberOfStems(); stemno++)
 117                                       {
 118                                             pPOS->AddStem (qSig->GetStem(stemno));
 119                                       }
 120
 121
 122                                       // Adjust affix frequencies for mentee.
 123                                       for (int affixno = 1; affixno <= qSig->Size(); ++affixno) {
 124                                               pPOS->IncrementPieceValue( qSig->GetPiece(affixno), qSig->GetNumberOfStems() );
 125                                               outf << "<td>" << qSig->GetPiece(affixno).Display() <<"</td><td>" <<qSig->GetNumberOfStems() <<"</td><td>";
 126
 127                                       }
 128                                       outf << EndTableRow ;
 129                               }// end of signo loop
 130
 131
 132                 // Write affix frequencies to log.
 133                 for (int m = 1; m <= pPOS->Size(); m++) {
 134                          outf << "<tr><td>" << pPOS->GetPiece(m).Display() <<"</td><td>"
 135                                  << pPOS->GetPieceValue(m)<<"</td></tr>";
 136                 }
 137         }
 138         outf <<"</table></body></html>";
 139 }