public static GisModel TrainModel(IEnumerable <string> files, int iterations, int cut, IEndOfSentenceScanner scanner)
{
var trainer = new GisTrainer();
foreach (var file in files)
{
using (var streamReader = new StreamReader(file))
{
ITrainingDataReader <string> dataReader = new PlainTextByLineDataReader(streamReader);
ITrainingEventReader eventReader = new SentenceDetectionEventReader(dataReader, scanner);
trainer.TrainModel(eventReader, iterations, cut);
}
}
return(new GisModel(trainer));
}
/// <summary>
/// Trains a model with a collection of input files with the following format:
/// [email protected] 0
/// mqsldkqsmlqsmdklqs@sdlsqjd 1
/// ...
/// The first line represents a valid email, the second an invalid.
/// </summary>
/// <param name="filePaths">The collection of file paths</param>
/// <param name="iterations">The number of iterations for the training</param>
/// <param name="cut">The cut for the training</param>
/// <returns>The trained GisModel</returns>
public static GisModel TrainModel(IEnumerable <string> filePaths, int iterations, int cut)
{
var trainer = new GisTrainer();
foreach (var file in filePaths)
{
using (var streamReader = new StreamReader(file))
{
ITrainingDataReader <string> dataReader = new PlainTextByLineDataReader(streamReader);
ITrainingEventReader eventReader = new InvalidEmailDetectionEventReader(dataReader);
trainer.TrainModel(eventReader, iterations, cut);
}
}
return(new GisModel(trainer));
}
public static GisModel TrainModel(IEnumerable<string> files, int iterations, int cut, IEndOfSentenceScanner scanner)
{
var trainer = new GisTrainer();
foreach (var file in files)
{
using (var streamReader = new StreamReader(file))
{
ITrainingDataReader<string> dataReader = new PlainTextByLineDataReader(streamReader);
ITrainingEventReader eventReader = new SentenceDetectionEventReader(dataReader, scanner);
trainer.TrainModel(eventReader, iterations, cut);
}
}
return new GisModel(trainer);
}
