/// <summary>
/// <see cref="ITextIndexes"/>の内容をtsvとして出力します。
/// </summary>
/// <param name="indexes"></param>
/// <param name="outputFilePath"></param>
private void WriteWordIndexStore(ITextIndexes indexes, string outputFilePath)
{
using (var stream = new StreamWriter(outputFilePath, append: false, encoding: Encoding.UTF8))
{
foreach (var entry in indexes.AllEntries())
{
stream.Write(entry.Text);
stream.Write("\t");
stream.Write(entry.Index);
stream.WriteLine();
}
stream.Flush();
}
}
public void Export(CntkTrainDataSet trainData)
{
// trainning_data.tsv
this.WriteTranInputData(trainData.Sentences);
// label
Console.WriteLine("creating label_index.tsv in folder : Data");
string labelFile = @"Data\labels_index.tsv";
this.WriteWordIndexStore(trainData.LabelIndexes, labelFile);
// words
Console.WriteLine("creating words_index.tsv in folder: Data");
ITextIndexes indexes = (trainData.WordSpace).WordIndexes;
if (indexes != null)
{
string wordsFilePath = @"Data\words_index.tsv";
WriteWordIndexStore(indexes, wordsFilePath);
}
}
/// <summary>
/// 単語インデックス辞書を用いて<see cref="OneHotWordVectorSpace"/>を構成します。
/// </summary>
/// <param name="wordIndexes"></param>
public static OneHotWordVectorSpace Load(ITextIndexes wordIndexes)
{
return(new OneHotWordVectorSpace(wordIndexes));
}
private OneHotWordVectorSpace(ITextIndexes wordIndexes)
{
this.WordIndexes = wordIndexes;
this.DimensionSize = this.WordIndexes.Count;
this.Zero = new ZeroWordVector(this.DimensionSize);
}
