/// <summary>
/// train with list of folder as array
/// </summary>
/// <param name="folderPathArr"></param>
private async Task<bool> Train(string[] folderPathArr)
{
_model = new LanguageModelBuilder();
const string progressTitle = "Please wait...";
const string progressTemplate = "Folder: {0}\nTraining {1}/{2} file(s) ({3}%)...";
var controller =
await _window.ShowProgressAsync(progressTitle, String.Format(progressTemplate, "", 0, 0, 0));
try
{
var sw = new Stopwatch();
foreach (var folderPath in folderPathArr)
{
sw.Restart();
var fileArr = Directory.GetFiles(folderPath);
//set initial progress
controller.SetMessage(String.Format(progressTemplate, folderPath, 0, fileArr.Length, 0));
await Task.Delay(1);
var cnt = 0;
var currentPercentage = 0;
foreach (var fileName in fileArr)
{
_model.Train(File.ReadAllText(fileName));
cnt++;
if (cnt*100/fileArr.Length <= currentPercentage) continue;
currentPercentage = cnt * 100 / fileArr.Length;
await Task.Delay(1);
controller.SetMessage(String.Format(progressTemplate, folderPath, cnt, fileArr.Length, currentPercentage));
controller.SetProgress((double)cnt / fileArr.Length);
}
//training with unknown word
_model.Train(LanguageModelBuilder.UnknownWordStr);
//marked as is trained
IsTrained = true;
_elapsedTime = (double)sw.ElapsedMilliseconds / 1000;
sw.Stop();
TrainingLog += String.Format("Trained folder {0} with {1} file(s). Elapsed Time: {2:##.###}s",
folderPath,
fileArr.Length,
_elapsedTime) + "\n";
TrainingLog += String.Format("Total words: {0}. Total distinct words: {1}.",
_model.WordCount,
_model.WordDict.Count) + "\n";
TrainingLog += String.Format("Total sentences: {0}.",
_model.SentenceCount) + "\n";
}
}
catch (Exception ex)
{
ShowMessage(ex.Message);
}
await controller.CloseAsync();
return true;
}
public void Test(string trainFolder, string testFolder)
{
var model = new LanguageModelBuilder();
Console.WriteLine("Training first with {0}", trainFolder);
//model.TrainByFolder(trainFolder, ShowLog);
Console.WriteLine("Test with {0}", testFolder);
//var perplexity = model.TestSourceByFolder(testFolder, ShowLog);
//Console.WriteLine(perplexity);
}
public MainWindowViewModel(MetroWindow window)
{
_window = window;
_model = new LanguageModelBuilder();
_service = new FolderService();
//assing commands
ChooseTrainingFolder1Cmd = new DelegateCommand(ChooseTrainingFolder1);
ChooseTrainingFolder2Cmd = new DelegateCommand(ChooseTrainingFolder2);
TrainBothSetCmd = new DelegateCommand(TrainBothSet);
GetAllWordsCmd = new DelegateCommand(GetAllWords);
TrainAndTestCmd = new DelegateCommand<string>(TrainAndTest);
GenSentencesCmd = new DelegateCommand(GenSentences);
//get defaul training folder
SetDefaultTrainingFolder();
//marked trained as false
IsTrained = false;
//guess word
GuessWords = new ObservableCollection<string> { "", "", "", "", "", "" };
PrevWords = new ObservableCollection<string> { "", "", "", "", "", "" };
_lastChange = DateTime.Now;
}
