generateDistanceMatrix.cc

Go to the documentation of this file.
00001 /***************************************
00002 *
00003 * program 
00004 *
00005 ****************************************
00006 *
00007 * modification history:
00008 *
00009 * HM 090399 created the documentation
00010 *
00011 ****************************************
00012 *
00013 * compiler defines used:
00014 *
00015 * QUERY
00016 *
00017 ****************************************/
00018 
00019 #include "../config.h"
00020 
00021 
00022 #include "libMRML/include/TID.h"
00023 #include <iostream>
00024 #include "libMRML/include/CRelevanceLevelList.h"
00025 #include "CInvertedFileAccessor.h"
00026 #include "CInvertedFileQuery.h"
00027 #include "libGIFTQuInvertedFile/include/CQNBestFullyWeighted.h"
00028 #include "libGIFTQuInvertedFile/include/CWFBestFullyWeighted.h"
00029 #include "libGIFTQuInvertedFile/include/CQNNoNormalization.h"
00030 #include "libGIFTQuInvertedFile/include/CQNEuclideanLengthSquare.h"
00031 #include "libGIFTQuInvertedFile/include/CQNSquareDFLogICFSum.h"
00032 #include "libGIFTQuInvertedFile/include/CQNMaxDocumentFrequency.h"
00033 #include "libGIFTQuInvertedFile/include/CWFBestProbabilistic.h"
00034 #include "libGIFTQuInvertedFile/include/CWFClassicalIDF.h"
00035 #include "libGIFTQuInvertedFile/include/CWFBinaryTerm.h"
00036 #include "libGIFTQuInvertedFile/include/CWFStandardTF.h"
00037 #include "libGIFTQuInvertedFile/include/CWFCoordinationLevel.h"
00038 #include "CWFNormal.h"
00039 #include "CWFColorHistogram.h"
00040 #include "CWFTextureHistogram.h"
00041 
00042 #include "CWFBlockOnly.h"
00043 #include "CWFColorBlockOnly.h"
00044 #include "CWFTextureBlockOnly.h"
00045 
00046 #include <algorithm>
00047 #include <cstdio>
00048 #include <stdlib.h>
00049 #include <string>
00050 
00051 
00052 TID gID;
00053 
00054 #define QUERY
00055 //#define W_PRUNING
00056 
00057 
00058 /****************************************
00059   compare by first element        
00060  ****************************************/
00061 bool sortByFirst(const pair<TID,float>& l,
00062      const pair<TID,float>& t){
00063   return 
00064     l.first
00065     >
00066     t.first;
00067 }
00068 
00069 
00070 /***************************************
00071 *
00072 * printListAsHTML - prints out the results as an html file to view it with a browser
00073 *
00074 ****************************************
00075 *
00076 * modification history
00077 *
00078 * 
00079 *
00080 ****************************************/
00081 
00082 void printListAsHTML(CRelevanceLevelList* inQueryResultList,
00083          string inQueryResultFileName){      
00084   if(inQueryResultList)
00085     {
00086       //an input-output-file
00087       ofstream lQueryResultFile(inQueryResultFileName.c_str(),
00088              fstream::in+fstream::out);
00089       if(lQueryResultFile)
00090   {
00091     lQueryResultFile << "<HTML>"
00092          << endl 
00093          << "<BODY>"
00094          << endl
00095          << "<TABLE>"
00096          << endl
00097          << "<TR>";
00098     
00099     int lCount=0;
00100     for(CRelevanceLevelList::const_iterator i=inQueryResultList->begin();
00101         i!=inQueryResultList->end() 
00102     && lCount<NUMBER_OF_IMAGES;
00103         i++,lCount++){
00104       
00105             /* HACK. Squire. 981030 */
00106             if (lCount == 20)
00107               lQueryResultFile << "</TR>" << endl << "</TABLE>" << endl
00108                                << "<!--" // Comment out those with rank > 50
00109                                << endl;
00110 
00111       if(!(lCount%5))
00112         lQueryResultFile << "</TR>"
00113              << endl
00114              << "<TR>";
00115       
00116       lQueryResultFile << "<TD>" 
00117            << endl;
00118       (*i).outputHTML(lQueryResultFile);  
00119       lQueryResultFile << "</TD>" 
00120            << endl;
00121       
00122     }
00123     
00124     lQueryResultFile << "</TR>"
00125          << endl
00126          << "</TABLE>"
00127          << endl 
00128                            << "-->" // Close the above comment
00129                            << endl
00130          << "</BODY>"
00131          << endl
00132          << "</HTML>";
00133   }
00134     }else
00135       cout << "failed" 
00136      << endl;
00137  
00138 }
00139 
00140 
00141 
00142 /***************************************
00143 *
00144 * main - main program, that reads the parameters and 
00145 *        executes the creation of the inverted file or the start of a query
00146 *
00147 ****************************************
00148 *
00149 * modification history
00150 *
00151 * 
00152 *
00153 ****************************************/
00154 
00155 main(int argc,
00156      char** argv){
00157 
00158   if(argc!=2){
00159     cout << "Usage: "
00160    << argv[0]
00161    << " <number_of_entries_to_skip>"
00162    << endl;
00163   }
00164   //
00165   const int lSkipNumber=atoi(argv[1]);
00166 
00167   
00168 
00169   cout << "Constructing Accessor..." 
00170        << flush;
00171 
00172   CInvertedFileAccessor lAccessor(INVERTED_FILE_LOCATION,
00173           OFFSET_FILE_LOCATION,
00174           FEATURE_FILE_LOCATION,
00175           FEATURE_DESCRIPTION_LOCATION);
00176   cout << "...after construction"
00177        << endl;
00178   
00179   
00180   cout << "Classic" << flush;
00181   CQueryNormalizer*   
00182     lDocumentNormalizer(new CQNNoNormalization(&lAccessor));
00183   CQueryNormalizer*   
00184     lQueryNormalizer(new CQNNoNormalization(&lAccessor));
00185   CWeightingFunction* lWeightingFunction=
00186     new CWFClassicalIDF(&lAccessor,
00187       lQueryNormalizer,
00188       lDocumentNormalizer);
00189   
00190   CRelevanceLevelList* lOutList;  
00191   
00192   cout << "Constructing Query-handler..."
00193    << flush;
00194   CInvertedFileQuery lQuery(lAccessor,
00195           *lWeightingFunction,
00196           *lQueryNormalizer,
00197           *lDocumentNormalizer);
00198 #ifdef W_PRUNING
00199   lQuery.activateBlockingFeatures();
00200   //  lQuery.blockFeatureGroup(COL_POS);
00201   //lQuery.blockFeatureGroup(GABOR_HST); // in an old version this was wrong...
00202   lQuery.blockFeatureGroup(GABOR_POS);
00203 #endif
00204 
00205   cout << "...after construction"
00206        << endl;
00207   
00208   
00209   list<TID> lAllIDs;
00210   
00211   lAccessor.getAllIDs(lAllIDs);
00212   
00213   lAllIDs.sort();
00214   
00215   {
00216     ofstream lTranslatorFile(DISTANCE_MATRIX_HOME
00217            //no comma<-> one string
00218            ".trans");
00219   int lCounter(0);
00220   for(list<TID>::const_iterator i=lAllIDs.begin();
00221       i!=lAllIDs.end();
00222       i++,
00223         lCounter++){
00224     lTranslatorFile << *i
00225         << " "
00226         << lCounter
00227         << endl;
00228   }
00229   }
00230   
00231   fstream lDistanceFile(DISTANCE_MATRIX_HOME,
00232        fstream::in+fstream::out);
00233 
00234   lDistanceFile.seekp(lAllIDs.size() * lSkipNumber * sizeof(float));
00235   //the main loop
00236   int lSkipCount=0;
00237   list<TID>::const_iterator i=lAllIDs.begin();
00238   for(;
00239       (i!=lAllIDs.end()) && (lSkipCount<lSkipNumber);
00240       i++,lSkipCount++){
00241     //do nothing
00242   };
00243   
00244   for(;
00245       i!=lAllIDs.end();
00246       i++){
00247 
00248     CRelevanceLevelList lRLL;
00249     lRLL.push_back(CRelevanceLevel(lAccessor.IDToURL(*i),1));
00250     
00251     cout << "The Query:"
00252    << lRLL.back().getURL()
00253    << endl;
00254     
00255     //here happens most of the calculation
00256     lOutList=lQuery.startQuery(lRLL,
00257              lAllIDs.size());
00258     {
00259       if(lOutList){
00260   // copy the distances into a list
00261   list<pair<TID,float> > lLineVector;
00262   
00263   for(CRelevanceLevelList::iterator j=lOutList->begin();
00264       j!=lOutList->end();
00265       j++){
00266     lLineVector.push_back(make_pair(lAccessor.URLToID(j->getURL()),
00267             j->getRelevanceLevel()));
00268   }
00269   lLineVector.sort();//sortByFirst)
00270   
00271     
00272   //
00273   int lWriteCount=0;
00274   //copy this list into a file
00275   int lOld=0;
00276   for(list<pair<TID,float> >::const_iterator j=lLineVector.begin();
00277       j!=lLineVector.end();
00278       j++){
00279     for(;
00280         lOld < j->first;
00281         lOld++){
00282       float lBuffer(0.0);
00283       lDistanceFile.write(&lBuffer,sizeof(lBuffer));
00284       lWriteCount++;
00285     }
00286     lOld++;
00287 
00288     cout << "["
00289          << j->first
00290          << ","
00291          << j->second
00292          << "]";
00293     
00294     float lDistance(j->second);
00295     
00296     lDistanceFile.write(&lDistance,sizeof(lDistance));
00297     lWriteCount++;
00298   }
00299   for(;
00300       lOld < lAllIDs.size();
00301       lOld++){
00302     float lBuffer(0.0);
00303     lDistanceFile.write(&lBuffer,sizeof(lBuffer));
00304     lWriteCount++;
00305   }
00306   lDistanceFile << flush;
00307   
00308   cout << endl
00309        << "Written: "
00310        << lWriteCount
00311        << "----------------------------------------Size: "
00312        << lLineVector.size()
00313        << endl;
00314       }
00315     }
00316     
00317     delete lOutList;
00318   }
00319 }
00320 
00321 
00322 
00323