Keep track of original index of document in MSet.

Store original index of document in MSet in FeatureVector.
VcamX · Jul 14, 2014 · 7e28b8f · 7e28b8f
1 parent 16ad87f
commit 7e28b8f
Show file tree

Hide file tree

Showing 3 changed files with 19 additions and 1 deletion.
diff --git a/xapian-letor/feature.cc b/xapian-letor/feature.cc
@@ -367,6 +367,8 @@ Feature::generate_feature_vector(const Xapian::MSetIterator & met_it_) {
 
     fvector.set_did( get_did(doc) );
 
+    fvector.set_index( mset_it_->get_rank() );
+
     vector<double> feature_values( get_features_num() );
     for (int i=0; i<get_features_num(); i++) {
         feature_values[i] = get_feature(features[i], mset_it_);

diff --git a/xapian-letor/feature_vector.cc b/xapian-letor/feature_vector.cc
@@ -70,6 +70,11 @@ FeatureVector::set_label(double label_) {
     label = label_;
 }
 
+void
+FeatureVector::set_index(Xapian::doccount index_) {
+    index = index_;
+}
+
 void
 FeatureVector::set_feature_values(vector<double> feature_values_) {
     feature_values = feature_values_;
@@ -90,6 +95,11 @@ FeatureVector::get_label() {
     return label;
 }
 
+Xapian::doccount
+FeatureVector::get_index() {
+    return index;
+}
+
 int
 FeatureVector::get_feature_num() {
     return feature_values.size();

diff --git a/xapian-letor/feature_vector.h b/xapian-letor/feature_vector.h
@@ -41,7 +41,7 @@ class XAPIAN_VISIBILITY_DEFAULT FeatureVector {
     double label;                   // the tagged relavance label
     double score;                   // the calculated score
     vector<double> fvals;           // feature values
-    double normalization_factor;    // the normalization factor
+    Xapian::doccount index;           // the original index in MSet
 
 public:
     FeatureVector();
@@ -58,6 +58,9 @@ class XAPIAN_VISIBILITY_DEFAULT FeatureVector {
 
     // Set the label
     void set_label(double label_);
+
+    // Set the original index in MSet
+    void set_index(Xapian::doccount index_);
 
     // Set feature values
     void set_feature_values(vector<double> feature_values_);
@@ -70,6 +73,9 @@ class XAPIAN_VISIBILITY_DEFAULT FeatureVector {
 
     // Get the label
     double get_label();
+
+    // Get the original index in MSet
+    Xapian::doccount get_index();
 
     // Get the number of features
     int get_feature_num();