Search and Evaluation

QtLing/Lexicon.cpp

@@ -16,6 +16,7 @@
 #include "SuffixCollection.h"
 #include "WordCollection.h"
 #include "Word.h"
+#include "evaluation.h"
 #include "cparse.h"
 
 CLexicon::CLexicon( CLexicon* lexicon, bool suffix_flag)
@@ -39,6 +40,8 @@ CLexicon::CLexicon( CLexicon* lexicon, bool suffix_flag)
     m_Hypothesis_map        = new QMap<QString, CHypothesis*>;
     m_entropy_threshold_for_stems = 1.2;
     m_parent_lexicon        = lexicon;
+    m_goldstandard          = NULL;
+    m_eval_parses      = NULL;
 
 //  This is part of an experiment.
     m_category_types["Words"]               = CT_word;
@@ -77,6 +80,7 @@ CLexicon::~CLexicon()
     delete m_PrefixSignatures;
     delete m_ParaSignatures;
     delete m_PassiveSignatures;
+    delete m_goldstandard;
 }
 
 CSignatureCollection* CLexicon::get_active_signature_collection(){
@@ -112,10 +116,56 @@ void CLexicon::clear_lexicon(){
     m_Hypotheses = new QList<CHypothesis*>;
 
 
+}
+
+// for gold standard
+// Return true if evaluation succeeded
+// Return false if it did not
+GoldStandard* CLexicon::new_goldstandard_from_xml(QString& file_name)
+{
+    m_goldstandard = new GoldStandard(file_name);
+    return m_goldstandard;
+}
+
+bool CLexicon::do_gs_evaluation()
+{
+    if (m_goldstandard == nullptr) {
+        qDebug() << 134 << "Lexicon.cpp: evaluation failed: GoldStandard not loaded";
+        return false;
+    }
+    bool evaluation_succeeded = m_goldstandard->evaluate(m_Words);
+    if (evaluation_succeeded) {
+        qDebug() << 139 << "Lexicon.cpp: evaluation completed";
+        return true;
+    } else return false;
+}
 
+EvalParses* CLexicon::new_eval_parses_from_txt(QString& file_name)
+{
+    m_eval_parses = new EvalParses(file_name);
+    return m_eval_parses;
+}
 
+void CLexicon::delete_eval_parses()
+{
+    delete m_eval_parses;
+    m_eval_parses = NULL;
 }
 
+bool CLexicon::do_gs_evaluation_on_eval_parses()
+{
+    if (m_goldstandard == NULL || m_eval_parses == NULL) {
+        qDebug() << 153 << "Lexicon.cpp: evaluation failed: GoldStandard or evaluation file not loaded";
+        return false;
+    }
+    bool evaluation_succeeded = m_goldstandard->evaluate(m_eval_parses);
+    if (evaluation_succeeded) {
+        qDebug() << 158 << "Lexicon.cpp: evaluation on imported parses completed";
+        return true;
+    } else return false;
+}
+
+
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 //linguistic methods
 /**
@@ -138,6 +188,12 @@ void CLexicon::dump_signatures_to_debug()
 
 }
 
+/* Crab_1:
+ * Used after MainWindow::read_dx1_file, which parses the dx1 file and
+ * stores words and their counts into CWordCollection object in Lexicon,
+ * and generates the SortedStringArrays.
+ *
+ */
 void CLexicon::Crab_1()
 {
     FindProtostems();
@@ -166,6 +222,9 @@ void CLexicon::Crab_1()
  * This is the first of the three initial parts of finding signatures.
  * This makes a cut at every point in a word where the successor frequency
  * is greater than 1.
+ * Taking the sorted string array as input, finds protoroots and stores them
+ * by modifying m_suffix_protostems_2 (for suffixes)
+ * or m_prefix_protostems_2 (for prefixes)
  */
 void CLexicon::FindProtostems()
 {   word_t          this_word, previous_word;
@@ -268,7 +327,9 @@ void CLexicon::FindProtostems()
 
 /*!
  * This is the second of the three initial parts of finding signatures.
- * This creates stem/affix pairs, which are put in a long list of "Parses".
+ * This creates stem/affix pairs, which are put in a long list of "Parses":
+ * QList<QPair<QString,QString>>* m_Parses
+ *
  */
 void CLexicon::CreateStemAffixPairs()
 {
@@ -377,7 +438,9 @@ void   CLexicon::assign_suffixes_to_stems(QString name_of_calling_function)
     stem_list *                 p_this_stem_list;
     //affix_set *                 this_ptr_to_affix_set(NULL);
     map_sigstring_to_suffix_set temp_stems_to_affix_set;
+    // Equivalent to QMap<QString, QList<QString>>
     map_sigstring_to_stem_list  temp_signatures_to_stems;
+    // Equivalent to QSet<QString>
     morph_set *                 pSet;
     CWord*                      pWord;
     m_ProgressBar->reset();
@@ -400,6 +463,7 @@ void   CLexicon::assign_suffixes_to_stems(QString name_of_calling_function)
     m_StatusBar->showMessage("Form signatures: 2. tentative signatures.");
 
     int count= 0;
+    // equivalent to QMap<QString, QSet<String>*>::iterator
     QMapIterator<QString, morph_set*>   stem_iter(temp_stems_to_affix_set);    // part 1
     while (stem_iter.hasNext())                                                // make a presignature for each stem.
     {    qApp->processEvents();
@@ -442,7 +506,9 @@ void   CLexicon::assign_suffixes_to_stems(QString name_of_calling_function)
         affix_list this_affix_list = this_signature_string.split("=");
         if (p_this_stem_list->size() >= MINIMUM_NUMBER_OF_STEMS)
         {
+            // put signature strings into m_Signatures
             if( m_SuffixesFlag) {
+                // CSignature* pSig;
                 pSig = *m_Signatures       << this_signature_string;
             } else {
                 pSig = *m_PrefixSignatures << this_signature_string;
@@ -453,6 +519,8 @@ void   CLexicon::assign_suffixes_to_stems(QString name_of_calling_function)
                 this_affix_t = affix_iter_2.next();
                 link_signature_and_affix(pSig,this_affix_t);
             }
+            // for each stem in the list of stems in the map of signatures,
+            // use the function link_signature_and_stem()
             stem_list_iterator stem_iter(*p_this_stem_list);
             while (stem_iter.hasNext()){
                 this_stem_t = stem_iter.next();
@@ -514,13 +582,17 @@ void CLexicon::link_signature_and_affix(CSignature * pSig, affix_t this_affix)
 }
 void CLexicon::link_signature_and_stem(stem_t this_stem_t , CSignature*  pSig,  QString this_signature_string)
 {
+    // add a CStem object into m_suffixal_stems/m_prefixal_stems:
+    //   - add CSignature pointer to that CStem object
+    //   - add CStem pointer to CSignature object
     CStem* pStem;
     QString this_affix, this_word;
     m_SuffixesFlag ?
             pStem = m_suffixal_stems->find_or_add(this_stem_t):
             pStem = m_prefixal_stems->find_or_add(this_stem_t);
     pStem->add_signature (pSig);
     pSig->add_stem_pointer(pStem);
+
     int stem_count = 0;
     affix_list this_affix_list = this_signature_string.split("=");
     QListIterator<suffix_t> affix_iter(this_affix_list);
@@ -535,7 +607,7 @@ void CLexicon::link_signature_and_stem(stem_t this_stem_t , CSignature*  pSig,
         }
         CWord* pWord = m_Words->get_word(this_word);
         if (!pWord){
-               qDebug() << this_word <<  "Error: this_word not found among words. Line 486" << this_stem_t  << this_affix << pSig->get_key() << this_signature_string;
+               qDebug() << this_word <<  "Error: this_word not found among words. Line 505" << this_stem_t  << this_affix << pSig->get_key() << this_signature_string;
         } else{
                stem_count += pWord->get_word_count();
                pWord->add_parse_triple(this_stem_t, this_affix, pSig->get_key());
@@ -546,6 +618,8 @@ void CLexicon::link_signature_and_stem(stem_t this_stem_t , CSignature*  pSig,
     }
     pStem->set_count(stem_count);
 }
+
+
 bool contains(QList<QString> * list2, QList<QString> * list1){
     for (int i=0; i < list1->length();i++){
         bool success = false;
@@ -787,8 +861,3 @@ void CLexicon::collect_parasuffixes()
     }
     m_ParaSuffixes->sort_by_count();
 };
-
-
-
-
-

QtLing/Lexicon.h

@@ -10,14 +10,15 @@
 #include <QStatusBar>
 #include "SignatureCollection.h"
 #include "Typedefs.h"
+#include "evaluation.h"
+
 class MainWindow;
 class CWordCollection;
 class CStemCollection;
 class CSuffixCollection;
 class CPrefixCollection;
 class QProgressBar;
 class CHypothesis;
-
 class CParse;
 
 //  part of an experiment:
@@ -139,9 +140,9 @@ class CLexicon
     CPrefixCollection *             m_Prefixes;
     CSignatureCollection *          m_Signatures;
     CSignatureCollection *          m_PrefixSignatures;
-    CWordCollection *               m_Compounds;
+    CWordCollection *               m_Compounds; // nothing done yet
     //QList<QPair<QString,QString>> * m_Parses;
-    QList<CParse*> *                 m_Parses;
+    QList<CParse*> *                 m_Parses; //
 
     QMap<QString,int>               m_Parse_map;
 //    QMap<QString, int>              m_suffix_protostems;
@@ -151,22 +152,27 @@ class CLexicon
     // with a particular proto-stem (i.e., a word-beginning). This replaces using a huge signature to store
     // that same information.
     QMap<QString, protostem*>        m_suffix_protostems_2;
-    QMap<QString, protostem*>        m_prefix_protostems_2;
+    QMap<QString, protostem*>        m_prefix_protostems_2; // temporary data structures for crab1
 
 
     bool                            m_SuffixesFlag;
     CLexicon*                       m_parent_lexicon;
 
+    // all of the possible continuations
+    // affixes that are "thrown out" in Crab2
     CSignatureCollection*           m_ParaSignatures;   /*!<  the information we have about stems which we have not yet integrated into a morphological system. */
     CSuffixCollection *             m_ParaSuffixes;
     CStemCollection *               m_ResidualStems;
     CSignatureCollection *          m_ResidualPrefixSignatures;
     CStemCollection *               m_StemsFromSubsignatures;
     CSignatureCollection*           m_Subsignatures;
+
+    // Finds the difference between signatures, e.g. {ed, es, er, e, ing} vs {d, s, r, NULL}
     QList<simple_sig_graph_edge*>   m_SigGraphEdgeList; /*!< the sig_graph_edges in here contain only one word associated with each. */
     lxa_sig_graph_edge_map          m_SigGraphEdgeMap;  /*!< the sig_graph_edges in here contain lists of words associated with them. */
     CSignatureCollection *          m_PassiveSignatures;  /*!< these signatures have stems one letter off from another signature. */
     CSignatureCollection *          m_SequentialSignatures; /*! signatures where one affix leads to another signature. */
+    // Generalizes repeating
     QList<CHypothesis*> *            m_Hypotheses;
     QMap<QString, CHypothesis*> *            m_Hypothesis_map;
 // add component 1
@@ -176,12 +182,29 @@ class CLexicon
 
     double                          m_entropy_threshold_for_stems;
 
+    // experiment for gold standard evaluation code
+    GoldStandard*                   m_goldstandard;
+    EvalParses*                     m_eval_parses;
+    // end of experiment
+
 public:
     CLexicon(CLexicon* parent_lexicon = NULL, bool suffix_flag = true);
 public:
 
     ~CLexicon();
 
+    // experiment for gold standard evaluation code
+    GoldStandard*                               get_goldstandard()          { return m_goldstandard; }
+    GoldStandard*                               new_goldstandard_from_xml(QString& file_name);
+    void                                        delete_goldstandard()       { delete m_goldstandard; m_goldstandard = NULL; }
+    bool                                        do_gs_evaluation();
+
+    EvalParses*                                 get_eval_parses()      { return m_eval_parses; }
+    EvalParses*                                 new_eval_parses_from_txt(QString& file_name);
+    void                                        delete_eval_parses();
+    bool                                        do_gs_evaluation_on_eval_parses();
+
+
     void                                        dump_signatures_to_debug();
     // accessors and protostems
     void                                        dump_suffixes(QList<QString>*);

QtLing/QtLing.pro

@@ -4,7 +4,7 @@
 #
 #-------------------------------------------------
 
-QT       += core gui
+QT       += core gui xml
 
 greaterThan(QT_MAJOR_VERSION, 4): QT += widgets
 
@@ -40,6 +40,14 @@ SOURCES += main.cpp\
     signature_graph.cpp \
     string_group.cpp \
     cparse.cpp \
+    lexicon_json.cpp \
+    evaluation.cpp \
+    evaluation_goldstandard.cpp \
+    evaluation_evalparses.cpp \
+    evaluation_parsemap.cpp \
+    mainwindow_find.cpp \
+    mainwindow_menubar.cpp \
+    mainwindow_actions.cpp \
     stringalignment.cpp \
     allosignatures.cpp
 
@@ -65,6 +73,9 @@ HEADERS  += mainwindow.h \
     mainwindow.h \
     string_group.h \
     cparse.h \
+    evaluation.h \
+    mainwindow_find.h \
+    mainwindow_menubar.h \
     stringalignment.h \
     allosignatures.h