/// <summary>
/// コンストラクタ
/// </summary>
/// <param name="text">テキスト</param>
public ParagraphData(string text)
{
Text = text;
TokenList = WordLogic.GetTokenList(text);
TokenTbl = WordLogic.GetBasicTokenTbl(text);
TokenTypeTbl = WordLogic.GetTokenTypeTbl(TokenList);
InfoRate = AnalyzeLogic.CalcInfoRate(TokenList);
}
public void 品詞を限定したトークンリストを取得できること(string testData_type, int testData_tokenNum)
{
var text = "形態素解析とは、文法的な情報の注記の無い自然言語のテキストデータから、対象言語の文法や、辞書と呼ばれる単語の品詞等の情報にもとづき、形態素の列に分割し、それぞれの形態素の品詞等を判別する作業である。";
var tokenList = WordLogic.GetTokenList(text);
var extractedTokenList = AnalyzeLogic.ExtractTokenType(tokenList, testData_type);
Assert.AreEqual(testData_tokenNum, extractedTokenList.Count);
foreach (var token in extractedTokenList)
{
Assert.AreEqual(testData_type, token.Type);
}
}
/// <summary>
/// メインメニュー：ファイル > 文章解析
/// </summary>
/// <param name="sender">送信元</param>
/// <param name="e">イベント</param>
private void mwnuSentenceAnalyzeToolStripMenuItem_Click(object sender, EventArgs e)
{
try
{
mTxtResult.Text = string.Empty;
var sb = new StringBuilder();
var ctrl = new WordCtrl();
ctrl.Analyze(mTxtTgtText.Text);
sb.AppendLine("====================");
sb.AppendFormat("情報量={0:#.#}[%]", ctrl.InfoRate * 100).AppendLine();
sb.AppendLine("====================");
foreach (var key in ctrl.TokenTypeTbl.Keys)
{
sb.AppendLine(string.Format("\t=== {0} ===", key));
var extractTokenTbl = AnalyzeLogic.ExtractTokenType(ctrl.TokenTbl, key);
foreach (var token in extractTokenTbl.Keys)
{
sb.AppendLine(string.Format("\t\t{0}: {1}",
token.Replace("\0", ""),
extractTokenTbl[token].Count()));
}
}
foreach (var paragraph in ctrl.ParagraphList)
{
sb.AppendLine("------------------------------");
sb.AppendLine(paragraph.Text);
sb.AppendFormat("情報量={0:#.#}[%]", paragraph.InfoRate * 100).AppendLine();
foreach (var key in paragraph.TokenTypeTbl.Keys)
{
sb.AppendLine(string.Format("\t=== {0} ===", key));
var extractTokenTbl = AnalyzeLogic.ExtractTokenType(paragraph.TokenTbl, key);
foreach (var token in extractTokenTbl.Keys)
{
sb.AppendLine(string.Format("\t\t{0}: {1}",
token.Replace("\0", ""),
extractTokenTbl[token].Count()));
//foreach(var word in extractTokenTbl[token])
//{
// sb.AppendLine(string.Format("\t\t\t{0}", word.Word));
//}
}
}
}
mTxtResult.Text = sb.ToString();
}
catch (Exception ex)
{
MessageBox.Show(ex.Message);
}
}
/// <summary>
/// 解析処理
/// </summary>
/// <param name="text">テキスト</param>
public void Analyze(string text)
{
ParagraphList.Clear();
var strParagraphList = ParagraphLogic.SplitParagraph(text);
foreach (var strParagraph in strParagraphList)
{
ParagraphList.Add(new ParagraphData(strParagraph));
}
TokenList = WordLogic.GetTokenList(text);
TokenTbl = WordLogic.GetBasicTokenTbl(text);
TokenTypeTbl = WordLogic.GetTokenTypeTbl(TokenList);
InfoRate = AnalyzeLogic.CalcInfoRate(TokenList);
}