/// <summary>
/// Vovabularyを生成する
/// </summary>
public static async Task <Vocabulary> Create(IEnumerable <string> sentences, CancellationToken token = default)
{
if (sentences == null)
{
throw new ArgumentException("sentencesがnullになっています。");
}
var morphemes = new List <Morpheme>();
foreach (var sentence in sentences)
{
var result = await MorphAnalyzerClient.AnalyzeAsync(sentence, token);
foreach (var item in result)
{
morphemes.Add(item);
}
}
return(new Vocabulary(morphemes));
}
/// <summary>
/// Bog Of Wordsを用いてベクトルに変換する
/// </summary>
public async Task <int[]> ConvertAsync(string sentence, CancellationToken token = default)
{
if (string.IsNullOrEmpty(sentence))
{
throw new ArgumentException("sentenceがNullもしくはEmptyになっています");
}
var morphemes = await MorphAnalyzerClient.AnalyzeAsync(sentence, token);
var vec = new int[_vocabulary.Count];
foreach (var morpheme in morphemes)
{
var index = _vocabulary.GetId(morpheme.Surface);
if (index != -1)
{
vec[index]++;
}
}
return(vec);
}
