/// <summary>
/// 将所给字符串写入文件，每行一个字符，共3列，3个特征：【字特征、词性特征、词边界特征】
/// </summary>
/// <param name="text">要写入文件的字符串</param>
/// <param name="path">保存路径</param>
public static void WriteSentenceToFile3(string text, string path)
{
var sw = File.CreateText(path);
var words = Preprocessor.Cut(text);
foreach (var item in words)
{
if (item.Word.Length == 1)
{
sw.WriteLine(item.Word[0] + "\t" + item.Flag + "\t" + "W");
}
else
{
sw.WriteLine(item.Word[0] + "\t" + item.Flag + "\t" + "B");
//每个字一行
for (int i = 1; i < item.Word.Length - 1; i++)
{
sw.WriteLine(item.Word[i] + "\t" + item.Flag + "\t" + "M");
}
sw.WriteLine(item.Word[item.Word.Length - 1] + "\t" + item.Flag + "\t" + "E");
}
}
/*
* for (int i = 0; i < text.Length; i++)
* {
* sw.WriteLine(text[i]);
* }*/
sw.Close();
}
/// <summary>
/// 将一个或多个带有命名实体标注的anns格式的文件转为一个bio标注的文件（字特征、词性特征、词边界特征）
/// </summary>
/// <param name="bioFile">要保存的文件名</param>
/// <param name="annsFiles">要转换的文件列表</param>
public static void ConvertAnnsToBio3(string bioFile, params string[] annsFiles)
{
if (annsFiles == null)
{
return;
}
var fs = new FileStream(bioFile, FileMode.Create);
var sw = new StreamWriter(fs, Encoding.UTF8);
List <MicroBlogCalendar.Model.Pair> tokens = null;
StreamReader sr;
for (int j = 0; j < annsFiles.Length; j++)
{
sr = File.OpenText(annsFiles[j]);
string line = null;
while ((line = sr.ReadLine()) != null)
{
if (string.IsNullOrEmpty(line))
{
sw.WriteLine();
continue;
}
var word = line.Split();
tokens = Preprocessor.Cut(word[0]);
switch (word[1])
{
case "S-Name": //专有名词
if (tokens.Count > 0)
{
//单个字组成词
if (tokens[0].Word.Length == 1)
{
sw.WriteLine(tokens[0].Word[0] + "\t" + tokens[0].Flag + "\t" + "W" + "\t" + "B-Name");
}
else
{
sw.WriteLine(tokens[0].Word[0] + "\t" + tokens[0].Flag + "\t" + "B" + "\t" + "B-Name");
for (int i = 1; i < tokens[0].Word.Length - 1; i++)
{
sw.WriteLine(tokens[0].Word[i] + "\t" + tokens[0].Flag + "\t" + "M" + "\t" + "I-Name");
}
sw.WriteLine(tokens[0].Word[tokens[0].Word.Length - 1] + "\t" + tokens[0].Flag + "\t" + "E" + "\t" + "I-Name");
}
for (int i = 1; i < tokens.Count; i++)
{
if (tokens[i].Word.Length == 1)
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "W" + "\t" + "I-Name");
}
else
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "B" + "\t" + "I-Name");
for (int k = 1; k < tokens[i].Word.Length - 1; k++)
{
sw.WriteLine(tokens[i].Word[k] + "\t" + tokens[i].Flag + "\t" + "M" + "\t" + "I-Name");
}
sw.WriteLine(tokens[i].Word[tokens[i].Word.Length - 1] + "\t" + tokens[i].Flag + "\t" + "E" + "\t" + "I-Name");
}
}
}
break;
case "S-Person": //人名
if (tokens.Count > 0)
{
//单个字组成词
if (tokens[0].Word.Length == 1)
{
sw.WriteLine(tokens[0].Word[0] + "\t" + tokens[0].Flag + "\t" + "W" + "\t" + "B-Person");
}
else
{
sw.WriteLine(tokens[0].Word[0] + "\t" + tokens[0].Flag + "\t" + "B" + "\t" + "B-Person");
for (int i = 1; i < tokens[0].Word.Length - 1; i++)
{
sw.WriteLine(tokens[0].Word[i] + "\t" + tokens[0].Flag + "\t" + "M" + "\t" + "I-Person");
}
sw.WriteLine(tokens[0].Word[tokens[0].Word.Length - 1] + "\t" + tokens[0].Flag + "\t" + "E" + "\t" + "I-Person");
}
for (int i = 1; i < tokens.Count; i++)
{
if (tokens[i].Word.Length == 1)
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "W" + "\t" + "I-Person");
}
else
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "B" + "\t" + "I-Person");
for (int k = 1; k < tokens[i].Word.Length - 1; k++)
{
sw.WriteLine(tokens[i].Word[k] + "\t" + tokens[i].Flag + "\t" + "M" + "\t" + "I-Person");
}
sw.WriteLine(tokens[i].Word[tokens[i].Word.Length - 1] + "\t" + tokens[i].Flag + "\t" + "E" + "\t" + "I-Person");
}
}
}
break;
case "S-Location": //地名
if (tokens.Count > 0)
{
//单个字组成词
if (tokens[0].Word.Length == 1)
{
sw.WriteLine(tokens[0].Word[0] + "\t" + tokens[0].Flag + "\t" + "W" + "\t" + "B-Location");
}
else
{
sw.WriteLine(tokens[0].Word[0] + "\t" + tokens[0].Flag + "\t" + "B" + "\t" + "B-Location");
for (int i = 1; i < tokens[0].Word.Length - 1; i++)
{
sw.WriteLine(tokens[0].Word[i] + "\t" + tokens[0].Flag + "\t" + "M" + "\t" + "I-Location");
}
sw.WriteLine(tokens[0].Word[tokens[0].Word.Length - 1] + "\t" + tokens[0].Flag + "\t" + "E" + "\t" + "I-Location");
}
for (int i = 1; i < tokens.Count; i++)
{
if (tokens[i].Word.Length == 1)
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "W" + "\t" + "I-Location");
}
else
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "B" + "\t" + "I-Location");
for (int k = 1; k < tokens[i].Word.Length - 1; k++)
{
sw.WriteLine(tokens[i].Word[k] + "\t" + tokens[i].Flag + "\t" + "M" + "\t" + "I-Location");
}
sw.WriteLine(tokens[i].Word[tokens[i].Word.Length - 1] + "\t" + tokens[i].Flag + "\t" + "E" + "\t" + "I-Location");
}
}
}
break;
case "S-Organization": //机构名
if (tokens.Count > 0)
{
//单个字组成词
if (tokens[0].Word.Length == 1)
{
sw.WriteLine(tokens[0].Word[0] + "\t" + tokens[0].Flag + "\t" + "W" + "\t" + "B-Organization");
}
else
{
sw.WriteLine(tokens[0].Word[0] + "\t" + tokens[0].Flag + "\t" + "B" + "\t" + "B-Organization");
for (int i = 1; i < tokens[0].Word.Length - 1; i++)
{
sw.WriteLine(tokens[0].Word[i] + "\t" + tokens[0].Flag + "\t" + "M" + "\t" + "I-Organization");
}
sw.WriteLine(tokens[0].Word[tokens[0].Word.Length - 1] + "\t" + tokens[0].Flag + "\t" + "E" + "\t" + "I-Organization");
}
for (int i = 1; i < tokens.Count; i++)
{
if (tokens[i].Word.Length == 1)
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "W" + "\t" + "I-Organization");
}
else
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "B" + "\t" + "I-Organization");
for (int k = 1; k < tokens[i].Word.Length - 1; k++)
{
sw.WriteLine(tokens[i].Word[k] + "\t" + tokens[i].Flag + "\t" + "M" + "\t" + "I-Organization");
}
sw.WriteLine(tokens[i].Word[tokens[i].Word.Length - 1] + "\t" + tokens[i].Flag + "\t" + "E" + "\t" + "I-Organization");
}
}
}
break;
case "S-Event": //事件
if (tokens.Count > 0)
{
//单个字组成词
if (tokens[0].Word.Length == 1)
{
sw.WriteLine(tokens[0].Word[0] + "\t" + tokens[0].Flag + "\t" + "W" + "\t" + "B-Event");
}
else
{
sw.WriteLine(tokens[0].Word[0] + "\t" + tokens[0].Flag + "\t" + "B" + "\t" + "B-Event");
for (int i = 1; i < tokens[0].Word.Length - 1; i++)
{
sw.WriteLine(tokens[0].Word[i] + "\t" + tokens[0].Flag + "\t" + "M" + "\t" + "I-Event");
}
sw.WriteLine(tokens[0].Word[tokens[0].Word.Length - 1] + "\t" + tokens[0].Flag + "\t" + "E" + "\t" + "I-Event");
}
for (int i = 1; i < tokens.Count; i++)
{
if (tokens[i].Word.Length == 1)
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "W" + "\t" + "I-Event");
}
else
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "B" + "\t" + "I-Event");
for (int k = 1; k < tokens[i].Word.Length - 1; k++)
{
sw.WriteLine(tokens[i].Word[k] + "\t" + tokens[i].Flag + "\t" + "M" + "\t" + "I-Event");
}
sw.WriteLine(tokens[i].Word[tokens[i].Word.Length - 1] + "\t" + tokens[i].Flag + "\t" + "E" + "\t" + "I-Event");
}
}
}
break;
case "S-Count":
if (tokens.Count > 0)
{
//单个字组成词
if (tokens[0].Word.Length == 1)
{
sw.WriteLine(tokens[0].Word[0] + "\t" + tokens[0].Flag + "\t" + "W" + "\t" + "B-Count");
}
else
{
sw.WriteLine(tokens[0].Word[0] + "\t" + tokens[0].Flag + "\t" + "B" + "\t" + "B-Count");
for (int i = 1; i < tokens[0].Word.Length - 1; i++)
{
sw.WriteLine(tokens[0].Word[i] + "\t" + tokens[0].Flag + "\t" + "M" + "\t" + "I-Count");
}
sw.WriteLine(tokens[0].Word[tokens[0].Word.Length - 1] + "\t" + tokens[0].Flag + "\t" + "E" + "\t" + "I-Count");
}
for (int i = 1; i < tokens.Count; i++)
{
if (tokens[i].Word.Length == 1)
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "W" + "\t" + "I-Count");
}
else
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "B" + "\t" + "I-Count");
for (int k = 1; k < tokens[i].Word.Length - 1; k++)
{
sw.WriteLine(tokens[i].Word[k] + "\t" + tokens[i].Flag + "\t" + "M" + "\t" + "I-Count");
}
sw.WriteLine(tokens[i].Word[tokens[i].Word.Length - 1] + "\t" + tokens[i].Flag + "\t" + "E" + "\t" + "I-Count");
}
}
}
break;
case "S-Time": //日期时间
if (tokens.Count > 0)
{
//单个字组成词
if (tokens[0].Word.Length == 1)
{
sw.WriteLine(tokens[0].Word[0] + "\t" + tokens[0].Flag + "\t" + "W" + "\t" + "B-Time");
}
else
{
sw.WriteLine(tokens[0].Word[0] + "\t" + tokens[0].Flag + "\t" + "B" + "\t" + "B-Time");
for (int i = 1; i < tokens[0].Word.Length - 1; i++)
{
sw.WriteLine(tokens[0].Word[i] + "\t" + tokens[0].Flag + "\t" + "M" + "\t" + "I-Time");
}
sw.WriteLine(tokens[0].Word[tokens[0].Word.Length - 1] + "\t" + tokens[0].Flag + "\t" + "E" + "\t" + "I-Time");
}
for (int i = 1; i < tokens.Count; i++)
{
if (tokens[i].Word.Length == 1)
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "W" + "\t" + "I-Time");
}
else
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "B" + "\t" + "I-Time");
for (int k = 1; k < tokens[i].Word.Length - 1; k++)
{
sw.WriteLine(tokens[i].Word[k] + "\t" + tokens[i].Flag + "\t" + "M" + "\t" + "I-Time");
}
sw.WriteLine(tokens[i].Word[tokens[i].Word.Length - 1] + "\t" + tokens[i].Flag + "\t" + "E" + "\t" + "I-Time");
}
}
}
break;
default: //非实体
if (tokens.Count > 0)
{
//单个字组成词
for (int i = 0; i < tokens.Count; i++)
{
if (tokens[i].Word.Length == 1)
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "W" + "\t" + "O");
}
else
{
sw.WriteLine(tokens[i].Word[0] + "\t" + tokens[i].Flag + "\t" + "B" + "\t" + "O");
for (int k = 1; k < tokens[i].Word.Length - 1; k++)
{
sw.WriteLine(tokens[i].Word[k] + "\t" + tokens[i].Flag + "\t" + "M" + "\t" + "O");
}
sw.WriteLine(tokens[i].Word[tokens[i].Word.Length - 1] + "\t" + tokens[i].Flag + "\t" + "E" + "\t" + "O");
}
}
}
break;
}
}
sr.Close();
}
sw.Close();
fs.Close();
}