xapian-letor/ranklist.cc

   1 #include <ranklist.h>
   2
   3 using namespace std;
   4 using namespace Xapian;
   5
   6 //Ranklist(const Xapian::MSet & mset,const Xapian::Database & db,const Xapian::Query & query)
   7 RankList::RankList()
   8 {
   9     /*map<Xapian::docid,double> letor_mset;
  10
  11     map<string,long int> coll_len;
  12     coll_len=collection_length(letor_db);
  13
  14     map<string,long int> coll_tf;
  15     coll_tf=collection_termfreq(letor_db,letor_query);
  16
  17     map<string,double> idf;
  18     idf=inverse_doc_freq(letor_db,letor_query);
  19
  20     int first=1;                //used as a flag in QueryLevelNorm module
  21
  22   //the above list will be mapped to an integer with its feature id.
  23
  24
  25
  26
  27     map< int, list<double> >::iterator norm_outer;
  28     list<double>::iterator norm_inner;
  29
  30
  31     List2 doc_ids;
  32
  33     for (Xapian::MSetIterator i = mset.begin(); i != mset.end(); i++) {
  34         Xapian::Document doc = i.get_document();
  35
  36         FeatureVector fv;
  37         fv.set_database(letor_db);
  38         fv.set_query(letor_query);
  39         std::map<int,double> fvals=fv.transform(doc);
  40
  41         add_feature_vector(fv);
  42
  43         if (first==1) {
  44             for (int j=1;j<20;j++) {
  45                 List1 l;
  46                 l.push_back(fvals[j]);
  47                 norm.insert(pair <int , list<double> > (j,l));
  48             }
  49             first=0;
  50         } else {
  51             norm_outer=norm.begin();
  52             int k=1;
  53             for (;norm_outer!=norm.end();norm_outer++) {
  54                 norm_outer->second.push_back(fvals[k]);
  55                 k++;
  56             }
  57         }
  58
  59         }
  60         norm = normalise(norm,norm_outer,norm_inner);
  61         */
  62 }
  63
  64 //Map3 normalise(Map3 norm,map< int, list<double> >::iterator norm_outer,list<double>::iterator norm_inner) {
  65 void
  66 RankList::normalise() {
  67
  68     /* this is the place where we have to normalize the norm and after that store it in the file. */
  69
  70 /*    if (!norm.empty()) {
  71         norm_outer=norm.begin();
  72         norm_outer++;
  73         int k=0;
  74         for (;norm_outer!=norm.end();++norm_outer) {
  75             k=0;
  76             double max= norm_outer->second.front();
  77             for (norm_inner = norm_outer->second.begin();norm_inner != norm_outer->second.end(); ++norm_inner) {
  78                 if (*norm_inner > max)
  79                     max = *norm_inner;
  80             }
  81             for (norm_inner = norm_outer->second.begin();norm_inner!=norm_outer->second.end();++norm_inner) {
  82                 if (max!=0)      // sometimes value for whole feature is 0 and hence it may cause 'divide-by-zero'
  83                     *norm_inner /= max;
  84                 k++;
  85             }
  86         }
  87
  88     }
  89     return norm;
  90     */
  91 }
  92
  93 void
  94 RankList::add_feature_vector(const Xapian::FeatureVector fv) {
  95     rl.push_back(fv);
  96 }
  97
  98 void
  99 RankList::set_qid(std::string qid1) {
 100     this->qid=qid1;
 101 }