From 84e3100081609453c0f0032b188e8078a8ee7d60 Mon Sep 17 00:00:00 2001
From: weirdwizardthomas <thomas.koristka@gmail.com>
Date: Sat, 28 Mar 2020 13:49:07 +0100
Subject: [PATCH] Finished relevance calculation
---
querying/CMakeLists.txt | 2 +-
querying/src/Computor.cpp | 12 ++---
querying/src/Document.cpp | 16 +++++++
querying/src/Document.h | 26 +++++++++++
querying/src/DocumentWeight.cpp | 16 -------
querying/src/DocumentWeight.h | 23 ----------
querying/src/inverted_index/InputParser.cpp | 3 +-
querying/src/inverted_index/InvertedIndex.cpp | 44 ++++++-------------
querying/src/inverted_index/InvertedIndex.h | 16 +++----
querying/src/main.cpp | 6 +--
querying/src/space/Query.cpp | 10 ++---
querying/src/space/Query.h | 6 +--
querying/src/space/Space.cpp | 5 ---
13 files changed, 84 insertions(+), 101 deletions(-)
create mode 100644 querying/src/Document.cpp
create mode 100644 querying/src/Document.h
delete mode 100644 querying/src/DocumentWeight.cpp
delete mode 100644 querying/src/DocumentWeight.h
diff --git a/querying/CMakeLists.txt b/querying/CMakeLists.txt
index 1d665ab..9114188 100644
--- a/querying/CMakeLists.txt
+++ b/querying/CMakeLists.txt
@@ -7,6 +7,6 @@ add_subdirectory(lib/SQLiteCpp)
SET(CMAKE_CXX_FLAGS "-g -Wall -pedantic -Wextra")
-add_executable(main src/main.cpp src/inverted_index/InvertedIndex.cpp src/inverted_index/InvertedIndex.h src/inverted_index/InputParser.cpp src/inverted_index/InputParser.h src/space/Space.cpp src/space/Space.h src/space/Query.cpp src/space/Query.h src/Computor.cpp src/Computor.h src/DocumentWeight.cpp src/DocumentWeight.h src/exceptions/Exceptions.h)
+add_executable(main src/main.cpp src/inverted_index/InvertedIndex.cpp src/inverted_index/InvertedIndex.h src/inverted_index/InputParser.cpp src/inverted_index/InputParser.h src/space/Space.cpp src/space/Space.h src/space/Query.cpp src/space/Query.h src/Computor.cpp src/Computor.h src/Document.cpp src/Document.h src/exceptions/Exceptions.h)
target_link_libraries(main SQLiteCpp)
diff --git a/querying/src/Computor.cpp b/querying/src/Computor.cpp
index ebbd268..68fffb5 100644
--- a/querying/src/Computor.cpp
+++ b/querying/src/Computor.cpp
@@ -8,7 +8,6 @@ using namespace std;
Computor::Computor(Space space, Query query) : space(std::move(space)), query(std::move(query)) {}
-
map<int, double> Computor::compute() {
map<int, double> results;
@@ -18,10 +17,10 @@ map<int, double> Computor::compute() {
int ID = nextID(); //get lowest ID
double result = 0;
- for (const auto &term: availableTerms) /*Go through all the remaining terms*/ {
+ for (const auto& term: availableTerms) /*Go through all the remaining terms*/ {
try {
- double queryWeight = query.terms.at(term);
double documentWeight = space.getInvertedIndexByKey(term).getDocumentWeightByID(ID);
+ double queryWeight = query.terms.at(term);
result += documentWeight * queryWeight;
}
catch (const IDNotFoundException &e) {//inverted index does not contain given ID
@@ -31,7 +30,8 @@ map<int, double> Computor::compute() {
availableTerms.erase(term); //exhaust term
}
}
- results[ID] = result;
+ if(result > query.threshold)
+ results[ID] = result;
}
return results;
@@ -40,8 +40,8 @@ map<int, double> Computor::compute() {
int Computor::nextID() {
int lowestID = INT_MAX;
- for (const auto &queryRecord: query.terms)
- lowestID = min(space[queryRecord.first].getLowestID(), lowestID);
+ for (const auto &term: availableTerms)
+ lowestID = min(space[term].getLowestID(), lowestID);
return lowestID;
}
diff --git a/querying/src/Document.cpp b/querying/src/Document.cpp
new file mode 100644
index 0000000..7bcebb2
--- /dev/null
+++ b/querying/src/Document.cpp
@@ -0,0 +1,16 @@
+//
+// Created by tomas on 3/24/20.
+//
+
+#include "Document.h"
+
+Document::Document(const int id, const double weight) : ID(id), weight(weight) {}
+
+int Document::getID() const {
+ return ID;
+}
+
+double Document::getWeight() const {
+ return weight;
+}
+
diff --git a/querying/src/Document.h b/querying/src/Document.h
new file mode 100644
index 0000000..aabf0d1
--- /dev/null
+++ b/querying/src/Document.h
@@ -0,0 +1,26 @@
+//
+// Created by tomas on 3/24/20.
+//
+
+#ifndef QUERYING_DOCUMENT_H
+#define QUERYING_DOCUMENT_H
+
+
+class Document {
+private:
+ int ID;
+ double weight;
+
+public:
+ Document(int id, double weight);
+
+ Document() = default;
+
+ int getID() const;
+
+ double getWeight() const;
+
+};
+
+
+#endif //QUERYING_DOCUMENT_H
diff --git a/querying/src/DocumentWeight.cpp b/querying/src/DocumentWeight.cpp
deleted file mode 100644
index 198ea45..0000000
--- a/querying/src/DocumentWeight.cpp
+++ /dev/null
@@ -1,16 +0,0 @@
-//
-// Created by tomas on 3/24/20.
-//
-
-#include "DocumentWeight.h"
-
-DocumentWeight::DocumentWeight(const int id, const double weight) : ID(id), weight(weight) {}
-
-int DocumentWeight::getID() const {
- return ID;
-}
-
-double DocumentWeight::getWeight() const {
- return weight;
-}
-
diff --git a/querying/src/DocumentWeight.h b/querying/src/DocumentWeight.h
deleted file mode 100644
index 299bab0..0000000
--- a/querying/src/DocumentWeight.h
+++ /dev/null
@@ -1,23 +0,0 @@
-//
-// Created by tomas on 3/24/20.
-//
-
-#ifndef QUERYING_DOCUMENTWEIGHT_H
-#define QUERYING_DOCUMENTWEIGHT_H
-
-
-class DocumentWeight {
-private:
- int ID;
- double weight;
-
-public:
- DocumentWeight(int id, double weight);
-
- int getID() const;
-
- double getWeight() const;
-};
-
-
-#endif //QUERYING_DOCUMENTWEIGHT_H
diff --git a/querying/src/inverted_index/InputParser.cpp b/querying/src/inverted_index/InputParser.cpp
index a26ac60..b885d4f 100644
--- a/querying/src/inverted_index/InputParser.cpp
+++ b/querying/src/inverted_index/InputParser.cpp
@@ -5,6 +5,7 @@
#include "InputParser.h"
#include <utility>
+#include <deque>
#include "../../lib/json.hpp"
using namespace std;
@@ -17,7 +18,7 @@ map<string, InvertedIndex> InputParser::getInvertedIndices() {
json root = loadJsonFromFile();
map<string, InvertedIndex> invertedIndices;
for (const auto &[term, documentIDs]: root.items()) {
- vector<DocumentWeight> weights;
+ deque<Document> weights;
for (const auto &[documentID, weight] : documentIDs.items())
weights.emplace_back(stoi(documentID), weight);
diff --git a/querying/src/inverted_index/InvertedIndex.cpp b/querying/src/inverted_index/InvertedIndex.cpp
index ed048ee..c9af044 100644
--- a/querying/src/inverted_index/InvertedIndex.cpp
+++ b/querying/src/inverted_index/InvertedIndex.cpp
@@ -4,50 +4,34 @@
using namespace std;
-InvertedIndex::InvertedIndex(deque<DocumentWeight> documentWeights)
- : documentWeights(move(documentWeights)) {}
-
+InvertedIndex::InvertedIndex(deque<Document> documentWeights)
+ : documents(move(documentWeights)) {}
double InvertedIndex::getDocumentWeightByID(int ID) {
+ Document document{};
- if (position + 1 == documentWeights.size())
+ if (documents.empty())
throw EndOfIndexException();
-
- while (documentWeights.front().getID() != ID) {
- documentWeights.pop_front();
- }
- return documentWeights.front().getID();
-
- for (auto it = documentWeights.begin() + position; it != documentWeights.end(); it++) {
- int documentID = (*it).getID();
-
- if (documentID < ID)
+ while ((document = documents.front()).getID() != ID) {
+ if (documents.empty()) //end of the index; index is exhausted
+ throw EndOfIndexException();
+ if (document.getID() > ID) // IDs are sorted asc, i.e. if doc ID is higher, then the ID for is not in index
throw IDNotFoundException();
-
- if (documentID == ID) //match!
- return (*it).getWeight();
+ documents.pop_front();
}
- throw IDNotFoundException();
+ documents.pop_front();
+ return document.getWeight();
}
-const DocumentWeight &InvertedIndex::operator[](size_t i) {
- return documentWeights[i];
+const Document &InvertedIndex::operator[](size_t i) {
+ return documents[i];
}
int InvertedIndex::getLowestID() const {
- return documentWeights.front().getID();
+ return documents.front().getID();
}
-void InvertedIndex::forward(int ID) {
- auto it = documentWeights.begin() + position;
- while (it != documentWeights.end() && (*it).getID() != ID) {
- if (ID < (*it).getID())
- return;
- position++;
- it++;
- }
-}
diff --git a/querying/src/inverted_index/InvertedIndex.h b/querying/src/inverted_index/InvertedIndex.h
index 677715e..c682307 100644
--- a/querying/src/inverted_index/InvertedIndex.h
+++ b/querying/src/inverted_index/InvertedIndex.h
@@ -4,7 +4,7 @@
#include <string>
#include <deque>
-#include "../DocumentWeight.h"
+#include "../Document.h"
/**
* @brief Data container of term and its inverted index list
@@ -14,7 +14,7 @@
class InvertedIndex {
private:
//Attributes-------------
- std::deque<DocumentWeight> documentWeights; /**<Inverted index list of documents & their weights in which the term appears */
+ std::deque<Document> documents; /**<Inverted index list of documents & their weights in which the term appears */
public:
@@ -23,23 +23,23 @@ public:
* Constructor
* @param documentWeights Inverted index list of documents in which the term appears, and their weights
*/
- InvertedIndex(std::deque<DocumentWeight> documentWeights);
+ explicit InvertedIndex(std::deque<Document> documentWeights);
/**
- * @brief Finds the @ref DocumentWeight object with a given @ref DocumentWeight::ID in @ref documentWeights
+ * @brief Finds the @ref Document object with a given @ref Document::ID in @ref documents
*
* Given an ID, looks through the container starting at @ref position.
- * Assumes that the items are sorted by @ref DocumentWeight::ID, ascending.
+ * Assumes that the items are sorted by @ref Document::ID, ascending.
*
- * @param ID @ref DocumentWeight::ID to be found.
+ * @param ID @ref Document::ID to be found.
* @throws IDNotFoundException if no Document is found or it had been processed already
- * @return DocumentWeight with @ref DocumentWeight::ID == ID
+ * @return Document with @ref Document::ID == ID
*/
double getDocumentWeightByID(int ID);
int getLowestID() const;
- const DocumentWeight &operator[](size_t i);
+ const Document &operator[](size_t i);
};
diff --git a/querying/src/main.cpp b/querying/src/main.cpp
index e1d3879..6356b01 100644
--- a/querying/src/main.cpp
+++ b/querying/src/main.cpp
@@ -11,9 +11,9 @@ int main() {
Space space(InputParser("../../data/persistence/dummy.json").getInvertedIndices());
Query query({
- {"forest", 0.5},
- {"mountain", 0.4},
- {"nature", 0.3}}, 0);
+ {"forest", 0.2},
+ {"mountain", 0.1},
+ {"nature", 0.8}}, 0.5);
auto res = Computor(space, query).compute();
for (const auto &[id, value]: res)
diff --git a/querying/src/space/Query.cpp b/querying/src/space/Query.cpp
index 6d92bfe..fc4a322 100644
--- a/querying/src/space/Query.cpp
+++ b/querying/src/space/Query.cpp
@@ -6,14 +6,14 @@
using namespace std;
-Query::Query(std::map<std::string, double> terms, double threshold)
- : terms(std::move(terms)),
- termsKeyset(getKeyset(terms)),
+Query::Query(std::map<std::string, double> t, double threshold)
+ : termsKeyset(getKeyset(t)),
+ terms(std::move(t)),
threshold(threshold) {}
-set<string> Query::getKeyset(const map<std::string, double> &terms) {
+set<string> Query::getKeyset(const map<std::string, double> &t) {
set<string> dummy;
- for (const auto &term: terms)
+ for (const auto &term: t)
dummy.emplace(term.first);
return dummy;
}
diff --git a/querying/src/space/Query.h b/querying/src/space/Query.h
index bda25cd..f724a70 100644
--- a/querying/src/space/Query.h
+++ b/querying/src/space/Query.h
@@ -14,8 +14,8 @@
class Query {
public:
//Attributes-------------
- const std::map<std::string, double> terms; /**<Terms and their weights in the query */
const std::set<std::string> termsKeyset; /**<Keyset of terms */
+ const std::map<std::string, double> terms; /**<Terms and their weights in the query */
const double threshold; /**<Acceptable result threshold */
//Methods----------------
@@ -24,9 +24,9 @@ public:
* @param terms Terms and their weights in the query
* @param threshold Acceptable result threshold
*/
- explicit Query(std::map<std::string, double> terms, double threshold);
+ explicit Query(std::map<std::string, double> t, double threshold);
- static std::set<std::string> getKeyset(const std::map<std::string, double> &terms);
+ static std::set<std::string> getKeyset(const std::map<std::string, double> &t);
};
diff --git a/querying/src/space/Space.cpp b/querying/src/space/Space.cpp
index f1c8fc2..e4bf9b5 100644
--- a/querying/src/space/Space.cpp
+++ b/querying/src/space/Space.cpp
@@ -17,10 +17,5 @@ InvertedIndex &Space::getInvertedIndexByKey(const string &key) {
return terms.at(key);
}
-void Space::forward(int ID, const std::map<std::string, double> &q) {
- for (const auto &term: q)
- terms.at(term.first).forward(ID);
-}
-
--
GitLab