/// <summary>
/// 获取包含根节点在内的单词数组
/// </summary>
/// <returns></returns>
public CoNLLWord[] GetWordArrWithRoot()
{
var words = new CoNLLWord[word.Length + 1];
words[0] = CoNLLWord.ROOT;
Array.Copy(word, 0, words, 1, word.Length);
return(words);
}
/// <summary>
///
/// </summary>
/// <param name="word"></param>
/// <param name="i"></param>
/// <param name="j"></param>
/// <returns></returns>
public static IEnumerable <string> GenerateUniContext(CoNLLWord[] word, int i, int j)
{
var context = new List <string>();
context.Add(word[i].NAME + '→' + word[j].NAME);
context.Add(word[i].POSTAG + '→' + word[j].POSTAG);
context.Add(word[i].NAME + '→' + word[j].NAME + (i - j));
context.Add(word[i].POSTAG + '→' + word[j].POSTAG + (i - j));
CoNLLWord wordBeforeI = i - 1 >= 0 ? word[i - 1] : CoNLLWord.NULL;
CoNLLWord wordBeforeJ = j - 1 >= 0 ? word[j - 1] : CoNLLWord.NULL;
context.Add(wordBeforeI.NAME + '@' + word[i].NAME + '→' + word[j].NAME);
context.Add(word[i].NAME + '→' + wordBeforeJ.NAME + '@' + word[j].NAME);
context.Add(wordBeforeI.POSTAG + '@' + word[i].POSTAG + '→' + word[j].POSTAG);
context.Add(word[i].POSTAG + '→' + wordBeforeJ.POSTAG + '@' + word[j].POSTAG);
return(context);
}
/// <summary>
/// 构造一个句子
/// </summary>
/// <param name="lines">行列表，每行存储一个Word的相关信息</param>
public CoNLLSentence(List <CoNllLine> lines)
{
var size = lines.Count;
word = new CoNLLWord[size];
for (int i = 0; i < lines.Count; i++)
{
word[i] = new CoNLLWord(lines[i]);
var head = int.Parse(lines[word[i].ID - 1].value[6]) - 1;
if (head != -1)
{
word[i].HEAD = word[head];
}
else
{
word[i].HEAD = CoNLLWord.ROOT;
}
}
}
/// <summary>
/// 获取一组单词中指定目标位置处单词的上下文
/// 从当前单词的前2单词处（如果不存在，则使用NULL代替）开始，上下文为 活动单词名称/词性 + 目标位置mark + 活动单词位置与目标位置的差
/// </summary>
/// <param name="word"></param>
/// <param name="index"></param>
/// <param name="mark"></param>
/// <returns></returns>
public static IEnumerable <string> GenerateSingleWordContext(CoNLLWord[] word, int index, string mark)
{
var context = new List <string>();
CoNLLWord w = null;
for (int i = index - 2; i < index + 2 + 1; i++)
{
if (i >= 0 && i < word.Length)
{
w = word[i];
}
else
{
w = CoNLLWord.NULL;
}
context.Add(w.NAME + mark + (i - index));
context.Add(w.POSTAG + mark + (i - index));
}
return(context);
}
