WIP: Fixed computation of cosine similarity

querying/src/calculation/Computor.cpp

@@ -13,15 +13,21 @@ Computor::Computor(Space space, Query query)
         : space(std::move(space)),
           query(std::move(query)) {}
-vector<pair<int, double>> Computor::compute(Database & database, int document_id) {
+vector<pair<int, double>> Computor::compute(Database & database) {
     vector<pair<int, double>> results;
-    map<string, double> currentDocument = space.getTermsAndWeightsByID(database, document_id);
+    map<string, double> computedDocument;
     availableTerms = query.termsKeyset;
+    double vectorQuerySize;
+    for (const auto & entry : query.terms)
+        vectorQuerySize += entry.second * entry.second;
+
     while (!availableTerms.empty()) {
         int ID = nextID(); //get lowest ID
-        double result = 0, denominator = 0, tmp = 0;
+        double result = 0, denominator = 0;
+        // hrozne zpomaluje beh programu (v radu sekund!!) --> vymyslet jine reseni...
+        computedDocument = space.getTermsAndWeightsByID(database, ID);
         for (const auto &term: availableTerms) /*Go through all the remaining terms*/ {
             try {
@@ -37,13 +43,10 @@ vector<pair<int, double>> Computor::compute(Database & database, int document_id
             }
         }
-        for (const auto & entry : query.terms)
+        for (const auto & entry : computedDocument)
             denominator += entry.second * entry.second;
         
-        for (const auto & entry : currentDocument)
-            tmp += entry.second * entry.second;
-
-        denominator = sqrt(denominator * tmp);
+        denominator = sqrt(denominator * vectorQuerySize);
         // Input should not be zero vector but if it is, do not divide and "just" return wrong result..
         if (denominator != 0)
             result = result / denominator;

querying/src/calculation/Computor.h

@@ -38,10 +38,10 @@ public:
     /**
      * @brief Computes relevancies of documents to the query, from results removes documents that have relevancies below @ref Query::threshold
-     * @param database, document_id Database connection and ID of current document (query)
+     * @param database Database connection
      * @return Vector of pairs with document IDs and their similarity to the query, sorted by similarity desc
      */
-    std::vector<std::pair<int, double>> compute(Database & database, int document_id);
+    std::vector<std::pair<int, double>> compute(Database & database);
 };

querying/src/calculation/Space.cpp

@@ -22,7 +22,7 @@ const map<string, double> Space::getTermsAndWeightsByID(Database & database, int
     map<string, double> terms;
     vector<string> dummy = database.getTermsByDocumentID(document_id);
-    for (string term : dummy) {
+    for (const string & term : dummy) {
         InvertedIndex tmp = getInvertedIndexByKey(term);
         terms[term] = tmp.getDocumentWeightByID(document_id);
     }

querying/src/ui/MainPage.cpp

@@ -76,7 +76,7 @@ void MainPage::displayDetail(Space space, Wt::WContainerWidget * container, int
   // threshold je nyni nastaven na -1 --> ve vysledku budou i uplne rozdilne dokumenty
   Query query(space.getTermsAndWeightsByID(database, document.id), -1);
-  auto result = Computor(space, query).compute(database, document.id);
+  auto result = Computor(space, query).compute(database);
   // dodelat proklikavani na zobrazene podobne dokumenty
   for (size_t i = 0; i < 10 && i < result.size(); i++) {