Optimized the CLM-Miner algorithm + Added Max Node Requirement + Docu…

…mentation

Took 1 hour 43 minutes

README.md

@@ -37,6 +37,8 @@ If neither `-output` and `-history` are provided then the script pretty-prints f
  8 : Failed to open output file
  9 : Bad min support number
 10 : Bad mfis output file
+11 : Bad max nodes count
+12 : Max nodes not specified
 ```
 
 ## Example

include/Graph.hpp

@@ -9,7 +9,7 @@ class Graph
 {
     struct Node {
       char label{};
-      uint occurrence{};
+      size_t occurrence{};
 
       bool operator<(const Node & rhs) const {
         return label < rhs.label;
@@ -27,14 +27,11 @@ class Graph
       NLOHMANN_DEFINE_TYPE_INTRUSIVE(Edge, from, to, extraNodes, occurrence);
     };
 
-    int row_length;
-    int N;
-    std::list<std::string> transactions;
-    std::list<nlohmann::json> jsonHistory;
+    size_t maxNodes = 1;
+    size_t maxRowSize = maxNodes * maxNodes + maxNodes;
     std::list<Node> nodes;
-    std::list<Edge> rawEdges;
-    std::map<std::string, int> itemsets;
-    std::map<char, std::vector<int>> CLM;
+    std::list<Edge> edges;
+    std::map<char, std::vector<size_t>> CLM;
 
     /**
      * \brief Checks if the edge exists, if so, increments the weight
@@ -46,21 +43,21 @@ class Graph
      */
     bool incrementIfRawEdgeExists(char fromNode, char toNode, const std::list<char> &extraNodes) noexcept;
 
-    /**
-     * \return The number of nodes in the graph
-     */
-    [[nodiscard]] int getNodesCount() const noexcept;
-
     /**
      * \brief Maps node to an integer postion to use for building CLM
      * \param node The node to map
      * \return The position of the node
      */
-    [[nodiscard]] int mapNodeToPosition(char node) const noexcept;
-    [[nodiscard]] char mapPostionToNode(int node) const noexcept;
+    [[nodiscard]] size_t mapNodeToPosition(char node) const noexcept;
+    [[nodiscard]] char mapPostionToNode(size_t node) const noexcept;
 
 public:
-    NLOHMANN_DEFINE_TYPE_INTRUSIVE(Graph, transactions, nodes, rawEdges, itemsets, CLM, jsonHistory);
+    NLOHMANN_DEFINE_TYPE_INTRUSIVE(Graph, nodes, edges, CLM);
+
+    explicit Graph(const size_t maxNodes): maxNodes(maxNodes) {}
+    Graph() = default;
+
+    void setMaxNodes(size_t maxNodes) noexcept;
 
     /**
      * \brief Processes the passed string into the graph generating new nodes and
@@ -69,16 +66,11 @@ class Graph
     void processTransaction(const std::string &str);
 
     /**
-     * \brief Processes the graph data to fill the CLM
-     */
-    void processCLM();
-
-    /**
-     * \brief Processes MFIs using the minSup
-     * \param minSup The minimum support count
-     * \return A list of MFIs that have support count above or equal to minSup
-     */
-    std::list<std::string> processMFIs(int minSup);
+ * \brief Processes FIs above the minimum support using the CLM Miner algorithm
+ * \param minSup The minimum support count
+ * \return A list of FIs that have support count above or equal to minSup after using the CLM Miner
+ */
+    std::list<std::string> useCLM_Miner(int minSup);
 
     /**
      * \brief Serializes all of the graph data excluding JSON history into a

main.cpp

@@ -65,6 +65,30 @@ int main(int argc, char **argv)
         }
     }
 
+    int maxNodes;
+    if (arguments.contains("-max-nodes"))
+    {
+        try
+        {
+            maxNodes = std::stoi(arguments["-max-nodes"]);
+
+            if (maxNodes < 1)
+            {
+                throw std::exception();
+            }
+        } catch (std::exception& _)
+        {
+            std::cout << "Max nodes is not a valid number!\n";
+            std::cout << _.what();
+
+            return 11;
+        }
+    } else
+    {
+        std::cout << "Max nodes not provided!\n";
+        return 12;
+    }
+
     std::vector<std::string> transactions;
     const std::string transactions_str = arguments["-transactions"];
     for (size_t offset = 0; offset < transactions_str.length();)
@@ -84,7 +108,7 @@ int main(int argc, char **argv)
         offset = pos + 1;
     }
 
-    Graph graph;
+    Graph graph(maxNodes);
 
     // Set up graph from image file if specified
     if (arguments.contains("-image"))
@@ -94,6 +118,7 @@ int main(int argc, char **argv)
         if (!fstream.is_open())
             return 6;
         graph = nlohmann::json::parse(fstream).get<Graph>();
+        graph.setMaxNodes(maxNodes);
     }
 
     // Process all transactions
@@ -102,9 +127,6 @@ int main(int argc, char **argv)
         graph.processTransaction(transaction);
     }
 
-    // Process the CLM after processing all transactions
-    graph.processCLM();
-
     // Output history to file
     if (arguments.contains("-history"))
     {
@@ -130,29 +152,30 @@ int main(int argc, char **argv)
 
     if (minSupport != -1)
     {
-        if (arguments.contains("-mfis-output"))
+        if (arguments.contains("-fis-output"))
         {
-            std::ofstream fstream(arguments["-mfis-output"]);
+            std::ofstream fstream(arguments["-fis-output"]);
             if (!fstream.is_open())
                 return 10;
 
-            fstream << nlohmann::json(graph.processMFIs(minSupport));
+            fstream << nlohmann::json(graph.useCLM_Miner(minSupport));
             fstream.flush();
             fstream.close();
         } else
         {
-            std::cout << "No MFIs output file specified\n";
+            std::cout << "No FIs output file specified\n";
         }
     }
 
-    if (!(arguments.contains("-history") && arguments.contains("-output") && arguments.contains("-mfis-output")))
+    if (!(arguments.contains("-history") && arguments.contains("-output") && arguments.contains("-fis-output")))
     {
+        const auto& FIs = graph.useCLM_Miner(minSupport);
         std::cout << graph.toString();
 
-        std::cout << "MFIs:\n";
-        for (const auto& mfi : graph.processMFIs(minSupport))
+        std::cout << "FIs:\n";
+        for (const auto& FI : FIs)
         {
-            std::cout << '\t' << mfi << '\n';
+            std::cout << '\t' << FI << '\n';
         }
     }