Esempio n. 1
0
        //Private Methods
        //Removing of unwanthed characters

        //Calculating final probability document belonging to category

        private double CalculatePCatDoc(CategoryTable tableConservative, double totalUniqueWords, double PCatParty, List <double> PCatPartyDoc)

        {
            double CatLogTotal = 0;

            for (int i = 0; i < fileContents.Count; i++)
            {
                PCatPartyDoc.Add(tableConservative.GetPWordCat(fileContents[i], totalUniqueWords));
            }
            if (PCatPartyDoc.Any())
            {
                PCatPartyDoc.Add(PCatParty);
            }
            else
            {
                PCatPartyDoc.Add(0);
                PCatPartyDoc.Add(PCatParty);
            }

            for (int i = 0; i < PCatPartyDoc.Count() - 1; i++)
            {
                CatLogTotal += PCatPartyDoc[i];
            }
            return(CatLogTotal);
        }
Esempio n. 2
0
        //Removing stopwords and combining lemmatizations

        //Public Methods
        public Classification(string conservativeDoc, string LabourDoc, string libDemConDoc, string master, string stopWordDoc, string lemmatizationDoc)
        {
            try
            {
                SetSourceDirectory();
                Console.WriteLine("Please enter the name of the .txt document you wish to predict.");
                Console.WriteLine("Name of included test documents: 'test1', 'test2', 'test3'");
                string FileDirectory = srcDirectory + Console.ReadLine() + ".txt";
                if (File.Exists(FileDirectory))
                {
                    ParseTextDocument(FileDirectory, stopWordDoc, lemmatizationDoc);

                    StreamReader sr = new StreamReader(srcDirectory + master);

                    double totalConservative = 0;
                    double totalLabour       = 0;
                    double totalLibDemCon    = 0;
                    double totalUniqueWords  = 0;

                    string[] tempMaster = sr.ReadToEnd().Split(' ', '\n', '\t');
                    for (int i = 0; i < tempMaster.Length; i++)
                    {
                        if (tempMaster[i] == "Conservative")
                        {
                            totalConservative = double.Parse(tempMaster[i + 1]);
                        }
                        if (tempMaster[i] == "Labour")
                        {
                            totalLabour = double.Parse(tempMaster[i + 1]);
                        }
                        if (tempMaster[i] == "LibDemCon")
                        {
                            totalLibDemCon = double.Parse(tempMaster[i + 1]);
                        }
                        if (tempMaster[i] == "Count")
                        {
                            totalUniqueWords = int.Parse(tempMaster[i + 1]);
                        }

                        sr.Close();
                    }
                    if (Math.Abs((totalConservative + totalLabour + totalLibDemCon)) < 1e-11)
                    {
                        Console.WriteLine("\nNo words were found in the network");
                        Console.ReadKey();
                        return;
                    }
                    double PCatConservative = Math.Log((totalConservative / (totalConservative + totalLabour + totalLibDemCon)));

                    double PCatLabour = Math.Log((totalLabour / (totalConservative + totalLabour + totalLibDemCon)));

                    double PCatConservativeLibDemCoalition = Math.Log((totalLibDemCon / (totalConservative + totalLabour + totalLibDemCon)));

                    tableConservative = new CategoryTable(srcDirectory + conservativeDoc, totalUniqueWords);
                    tableLabour       = new CategoryTable(srcDirectory + LabourDoc, totalUniqueWords);
                    tableLibDem       = new CategoryTable(srcDirectory + libDemConDoc, totalUniqueWords);

                    double consLogTotal      = CalculatePCatDoc(tableConservative, totalUniqueWords, PCatConservative, PCatConDoc);
                    double LabourLogTotal    = CalculatePCatDoc(tableLabour, totalUniqueWords, PCatLabour, PCatLabourDoc);
                    double libDemConLogTotal = CalculatePCatDoc(tableLibDem, totalUniqueWords, PCatConservativeLibDemCoalition, PCatConservativeConseravtiveCoalitionDoc);

                    if (consLogTotal > LabourLogTotal && consLogTotal > libDemConLogTotal)
                    {
                        Console.WriteLine("\nMost Likely: Conservative");
                    }
                    else if (LabourLogTotal > consLogTotal && LabourLogTotal > libDemConLogTotal)
                    {
                        Console.WriteLine("\nMost Likely: Labour");
                    }
                    else if (libDemConLogTotal > consLogTotal && libDemConLogTotal > LabourLogTotal)
                    {
                        Console.WriteLine("\nMost Likely: Liberal Democrats / Conservative Coalition");
                    }
                    else
                    {
                        Console.WriteLine("\nCannot Determine Party!");
                    }

                    Console.WriteLine("\nConservative                             : LogE(Probability) = " + consLogTotal);
                    Console.WriteLine("Labour                                   : LogE(Probability) = " + LabourLogTotal);
                    Console.WriteLine("Conservative/Liberal Democrats Coalition : LogE(Probability) = " + libDemConLogTotal);
                    Console.WriteLine("\nNB: The unknown document is predicted by the most positive LogE(Probability).");
                    Console.WriteLine("\nPress any KEY to continue.");

                    Console.ReadKey();
                }
                else
                {
                    Console.Clear();
                    Console.WriteLine(FileDirectory + " Doesn't exist!");
                    Console.WriteLine("Please enter the name of a valid .txt file.");
                    Console.WriteLine("Press any key to restart.");
                    Console.WriteLine("Or Press ESC to exit.");

                    if (Console.ReadKey().Key == ConsoleKey.Escape)
                    {
                        Environment.Exit(0);
                    }
                }
            }
            catch (Exception e)
            {
                ExeptionMethod(e);
            }
        }