WordCollection.h

   1 // Mini-lexicon’s collection of words
   2 // Copyright © 2009 The University of Chicago
   3 #ifndef WORDCOLLECTION_H
   4 #define WORDCOLLECTION_H
   5
   6 // See the CMiniLexicon class in MiniLexicon.h for an overview of
   7 // suffix/signature-based discovery of morphology.
   8
   9 #include "CollectionTemplate.h"
  10 #include <QString>
  11 #include "StemListViewItem.h"
  12 #include "Stem.h"
  13 #include "AffixLocation.h"
  14 template<class K, class V> class QMap;
  15 namespace linguistica { namespace ui { class status_user_agent; } }
  16 class CSignatureCollection;
  17 class CPhoneCollection;
  18
  19 class CWordCollection : public TCollection<CStem> {
  20 public:
  21         enum eAffixLocation m_AffixLocation;
  22         enum CWordListViewItem::display_mode m_DisplayMode;
  23         CPhoneCollection* m_Phones_Tier1;
  24         CPhoneCollection* m_Phones_Tier2;
  25         CPhoneCollection* m_Phones_Tier1_Skeleton;
  26
  27         double m_PhonologicalContent_Unigram;
  28         double m_PhonologicalContent_Bigram;
  29         double m_Tier2_LocalMI_Score;
  30         double m_Tier2_DistantMI_Score;
  31
  32         /// includes Tier1 bigram content
  33         double m_LocalMI_TotalBoltzmannScore;
  34         double m_LocalMI_Plog;
  35         /// includes Tier1 bigram content
  36         double m_DistantMI_TotalBoltzmannScore;
  37         double m_DistantMI_Plog;
  38
  39         double m_MyZ_Local;
  40         double m_MyZ_Distant;
  41         CParse* m_Vowels;
  42 public:
  43         // construction/destruction.
  44
  45         explicit CWordCollection(CMiniLexicon* lexicon = 0);
  46         ~CWordCollection();
  47
  48         // disable copy
  49 private:
  50         CWordCollection(const CWordCollection& x);
  51         CWordCollection& operator=(const CWordCollection& x);
  52 public:
  53         CStem* operator<<(CStem*);
  54         CStem* operator<<(CStem&);
  55         CStem* operator<<(CStringSurrogate&);
  56         CStem* operator<<(CParse*);
  57         CStem* operator<<(QString);
  58         void AddPointer(CStem*);
  59         CStem* AddToCollection(CParse&);
  60         CStem* AddToCollection(CStringSurrogate&);
  61
  62         void Empty();
  63         void RemoveAll();
  64         bool Remove(CStem*);    ///< doesn't delete CStem*
  65         bool RemoveMember(CStem*);      ///< deletes CStem*
  66         bool RemoveMember(CStringSurrogate&);   ///< deletes CStem*
  67         bool RemoveMember(CStringSurrogate&, bool);
  68         void DeleteMarkedMembers();
  69
  70         // Accessors
  71         void AssignSignatureFromStemsAffixPointer(enum eAffixLocation);
  72         int HowManyAreAnalyzed(int& HowManyNotAnalyzed,
  73                 linguistica::ui::status_user_agent& status_display);
  74
  75         void OutputWords(QString filename, QMap<QString, QString>* filter);
  76
  77         void OutputWordsForTesting(QString);
  78         CPhoneCollection* GetPhones();
  79         CPhoneCollection* GetPhones_Tier2();
  80         CPhoneCollection* GetPhones_Tier1_Skeleton();
  81         // Mutators
  82         void ReadWordFile(QString);
  83         void PredecessorFreq1(CStemCollection*,
  84                 CPrefixCollection*, CSignatureCollection*,
  85                 enum eSuccessorFrequencyMode, int);
  86         void SuccessorFreq1(CStemCollection*,
  87                 CSuffixCollection*, CSignatureCollection*,
  88                 enum eSuccessorFrequencyMode, int);
  89         void FindAllWordNeighbors(CLexicon*);
  90         //Phonology
  91         void CountPhonesAndBiphones(enum eTier tier);
  92         void DoPhonology();
  93         void CreateCVTemplate();
  94
  95         void SplitPhonologyToTiers(enum CStem::ePhonologySplitType leave_slot);
  96         void CreatePhonologyFromOrthography();
  97         void ComputeProbabilitiesOfWords();
  98         void GetPhonologyTierInfoForGraphOfWords();
  99         double ComputeZeta();
 100         void ComputeBoltzmannProbabilities();
 101
 102         double ComputeZ();
 103         double ComputeZStar();  ///< Field method on tier 2
 104
 105         double GetPhonologicalContentTier1Bigrams()
 106                 { return m_PhonologicalContent_Bigram; }
 107         double GetPhonologicalContentUnigrams()
 108                 { return m_PhonologicalContent_Unigram; }
 109         double GetZ_Local() { return m_MyZ_Local; }
 110         double GetZ_Distant() { return m_MyZ_Distant; }
 111
 112         double GetTier2_LocalMI_Score() { return m_Tier2_LocalMI_Score; }
 113         double GetLocalMI_Plog() { return m_LocalMI_Plog; }
 114         double GetTier2_DistantMI_Score() { return m_Tier2_DistantMI_Score; }
 115         double GetDistantMI_Plog() { return m_DistantMI_Plog; }
 116 };
 117
 118 #endif // WORDCOLLECTION_H