/// <summary>
/// Add a training document to the network
/// </summary>
/// <param name="trainingDoc">The training doc to add to the network</param>
public void AddTrainingDoc(TrainingDoc trainingDoc)
{
Category cat = trainingDoc.Category;
// Get partyData of same category as the TrainingDoc
PartyData partyData = Data.Where(x => x.GetCategory() == cat).First();
partyData.IncrementDocCount();
// Recalculate prior probabilities as document count has changed
Dictionary <Category, double> priorProbabilities = _calculator.GetPriorProbabilities(Data);
foreach (KeyValuePair <Category, double> kvp in priorProbabilities)
{
if (kvp.Key != Category.NONE)
{
PartyData party = Data.Where(x => x.GetCategory() == kvp.Key).FirstOrDefault();
party.SetProbability(kvp.Value);
}
}
// Merge the existing word set with the new document word set
partyData.SetWords(MergeWords(partyData.Words, trainingDoc.Words));
// Merge the existing ngram set with the new document ngram set
partyData.SetNGrams(MergeWords(partyData.NGrams, trainingDoc.NGrams));
// Recalculate relative frequencies as training set has been modified
_calculator.GetRelativeFrequencies(partyData.Words);
_calculator.GetRelativeFrequencies(partyData.NGrams);
}
/// <summary>
/// Assign the party to the document
/// </summary>
/// <param name="fileName">The name of the file to assign the party to</param>
/// <returns></returns>
private TrainingDoc AssignDocumentParty(string fileName)
{
ConsoleKey choiceInput;
int choice;
Console.Clear();
TrainingDoc td = new TrainingDoc();
Console.WriteLine("Choose the party this document belongs to:\n");
Dictionary <int, Category> categoryPairs = new Dictionary <int, Category>()
{
{ 1, Category.CONSERVATIVE }, { 2, Category.COALITION }, { 3, Category.LABOUR }
};
foreach (KeyValuePair <int, Category> pair in categoryPairs)
{
Console.WriteLine($"{pair.Key}. {pair.Value}");
}
Console.Write("Enter an option: ");
choiceInput = Console.ReadKey().Key;
switch (choiceInput)
{
case ConsoleKey.D1:
td = new TrainingDoc(fileName, Category.CONSERVATIVE);
break;
case ConsoleKey.D2:
td = new TrainingDoc(fileName, Category.COALITION);
break;
case ConsoleKey.D3:
td = new TrainingDoc(fileName, Category.LABOUR);
break;
case ConsoleKey.Escape:
return(null);
default:
Console.WriteLine("Invalid Input!");
break;
}
return(td);
}
/// <summary>
/// Train the network with new data
/// </summary>
private void Train()
{
StartTraining:
string docChoice;
ConsoleKey choiceInput;
int choice;
bool validInput = false;
TrainingDoc t = new TrainingDoc();
do
{
Console.Clear();
Console.WriteLine("Choose a document to train the network with\n");
string[] docs = Directory.GetFiles("TrainingDocs");
Dictionary <int, string> trainingDocPairs = new Dictionary <int, string>();
Dictionary <int, string> trainingDocPairDisplay = new Dictionary <int, string>();
for (int i = 0; i < docs.Count(); i++)
{
trainingDocPairs[i + 1] = docs[i];
trainingDocPairDisplay[i + 1] = docs[i].Replace("TrainingDocs\\", "");
}
foreach (KeyValuePair <int, string> pair in trainingDocPairDisplay)
{
Console.WriteLine($"{pair.Key}. {pair.Value}");
}
Console.Write("Choose an option (Type 00 to go back): ");
docChoice = Console.ReadLine();
if (docChoice == "00")
{
return;
}
try
{
choice = Convert.ToInt32(docChoice);
if (trainingDocPairs.TryGetValue(choice, out string value))
{
validInput = true;
t = AssignDocumentParty(value);
if (t == null)
{
goto StartTraining;
}
}
else
{
Console.WriteLine("Invalid option chosen!");
Console.ReadLine();
}
}
catch (FormatException e)
{
Console.WriteLine(e.Message + " Try a number instead!");
Console.ReadLine();
}
} while (!validInput);
_network.AddTrainingDoc(t);
Console.WriteLine("Trained Network with \"{0}\"", t.FileName);
Console.ReadLine();
Console.Clear();
SavePrompt:
Console.WriteLine("Save the network to file?");
Console.WriteLine("1. Yes\n2. No\nHit enter after you have made your choice");
choiceInput = Console.ReadKey().Key;
switch (choiceInput)
{
case ConsoleKey.D1:
_network.Save();
Classify();
break;
case ConsoleKey.D2:
goto StartTraining;
default:
goto SavePrompt;
}
return;
}
