public static void PatternMatch(List <NS> nss, List <Vertex> vertices, WordNet wordnet_op, WordNet wordnet_all)
{
var sb = new StringBuilder(nss.Count);
for (int i = 0; i < nss.Count; i++)
{
sb.Append(nss[i].ToString());
}
var patternStr = sb.ToString();
var vertexArr = vertices.ToArray();
trie.Match(patternStr, (begin, end, keyword) =>
{
var sbName = new StringBuilder();
for (int i = begin; i < end; i++)
{
sbName.Append(vertexArr[i].realWord);
}
var name = sbName.ToString();
if (IsBadCase(name))
{
return;
}
int offset = 0;
for (int i = 0; i < begin; i++)
{
offset += vertexArr[i].realWord.Length;
}
wordnet_op.Insert(offset, new Vertex(TAG_PLACE, name, ATTRIBUTE, CoreDictionary.NS_WORD_ID), wordnet_all);
});
}
public static void PatternMatch(List <NT> nts, List <Vertex> vertices, WordNet wordNetOptimum, WordNet wordNetAll)
{
var sb = new StringBuilder(nts.Count);
for (int i = 0; i < nts.Count; i++)
{
sb.Append(nts[i].ToString());
}
var patternStr = sb.ToString();
var vertexArr = vertices.ToArray();
_trie.Match(patternStr, (begin, end, keyword) =>
{
var sbName = new StringBuilder();
for (int i = begin; i < end; i++)
{
sbName.Append(vertexArr[i].realWord);
}
var name = sbName.ToString();
if (IsBadCase(name))
{
return; // 对一些basecase 做出调整
}
int offset = 0;
for (int i = 0; i < begin; i++)
{
offset += vertexArr[i].realWord.Length;
}
wordNetOptimum.Insert(offset, new Vertex(TAG_GROUP, name, ATTRIBUTE, CoreDictionary.NT_WORD_ID), wordNetAll);
});
}
public static void Recognition(List <Vertex> vertices, WordNet wordNetOptimum, WordNet wordNetAll)
{
var sb = new StringBuilder();
int appendTimes = 0; // stringbuilder 附加次数
int line = 1; // 行号，跳过起始辅助节点
int activeLine = 1; //
for (int i = 1; i < vertices.Count; i++) // 遍历节点，跳过起始辅助节点
{
var vertex = vertices[i];
if (appendTimes > 0) // 已经附加过
{
// 如果顶点词性为音译人名，或者音译人名词典包含顶点字符串值
if (vertex.GuessNature() == Nature.nrf || TranslatedPersonDictionary.ContainsKey(vertex.realWord))
{
sb.Append(vertex.realWord);
++appendTimes;
}
else
{
// 识别结束
if (appendTimes > 1) // 附加两次才算是一个完整的音译人名吗？
{
wordNetOptimum.Insert(activeLine, new Vertex(TAG_PEOPLE, sb.ToString(), new WordAttr(Nature.nrf), CoreDictionary.NR_WORD_ID), wordNetAll);
}
sb.Clear();
appendTimes = 0;
}
}
else // 尚未附加过
{
if (vertex.GuessNature() == Nature.nrf || TranslatedPersonDictionary.ContainsKey(vertex.realWord))
{
sb.Append(vertex.realWord);
++appendTimes;
activeLine = line; // 第一次附加，记录活跃行号
}
}
line += vertex.realWord.Length; // 更新下一个顶点的行号
}
}
/// <summary>
/// 模式匹配
/// </summary>
/// <param name="nrs">确定的标注序列</param>
/// <param name="vertexs">原始的未加角色标注的序列</param>
/// <param name="wordNetOptimum"></param>
/// <param name="wordNetAll"></param>
public static void PatternMatch(List <NR> nrs, List <Vertex> vertexs, WordNet wordNetOptimum, WordNet wordNetAll)
{
var sb = new StringBuilder(nrs.Count); // 存储 NR 的枚举模式串
var preNR = NR.A;
bool backUp = false;
int index = 0;
for (int i = 0; i < nrs.Count; i++, index++)
{
var cur = vertexs[index];
switch (nrs[i])
{
case NR.U: // 人名上文和姓成词， 比如： 这里【有关】天培的壮烈
if (!backUp) // 如果尚未备份，则备份一下
{
vertexs = new List <Vertex>(vertexs);
backUp = true;
}
sb.Append(NR.K.ToString());
sb.Append(NR.B.ToString());
preNR = NR.B;
var nowK = cur.realWord.Substring(0, cur.realWord.Length - 1); // 人名的上文，参见上面 “有关”的“有”
var nowB = cur.realWord.Substring(cur.realWord.Length - 1); // 最后一个字表示姓氏，单独提取出来，参见上面 “有关”的“关”
// 因为匹配到人名前缀与人名合在一个节点里面，将当前节点拆分
vertexs[index] = new Vertex(nowK);
vertexs.Insert(++index, new Vertex(nowB));
continue;
case NR.V: // 人名末与下文成词，比如：“龚学平等领导”中的“龚学平等”，“邓颖超生前”中的“邓颖超生”
if (!backUp)
{
vertexs = new List <Vertex>(vertexs);
backUp = true;
}
if (preNR == NR.B)
{
sb.Append(NR.E.ToString()); // BE
}
else
{
sb.Append(NR.D.ToString()); // CD
}
sb.Append(NR.L.ToString());
var nowED = cur.realWord.Substring(cur.realWord.Length - 1); // 提取最后一个字
var nowL = cur.realWord.Substring(0, cur.realWord.Length - 1); //
vertexs[index] = new Vertex(nowL); // 1.
vertexs.Insert(++index, new Vertex(nowED)); // 2. 这两行与原java代码执行顺序相反
continue;
default:
sb.Append(nrs[i].ToString());
break;
}
preNR = nrs[i];
}
var patternStr = sb.ToString(); // 所有节点连接起来形成的模式串
var wordArr = vertexs.ToArray();
var offsetArr = new int[wordArr.Length];
offsetArr[0] = 0;
for (int i = 1; i < wordArr.Length; i++)
{
offsetArr[i] = offsetArr[i - 1] + wordArr[i - 1].realWord.Length;
}
_trie.Match(patternStr, (begin, end, value) =>
{
var sbName = new StringBuilder();
for (int i = begin; i < end; i++)
{
sbName.Append(wordArr[i].realWord);
}
var name = sbName.ToString();
switch (value)
{
case NRPattern.BCD:
if (name[0] == name[2])
{
return; // 认为姓和最后一个名不可能相等
}
break;
}
if (IsBadCase(name))
{
return;
}
wordNetOptimum.Insert(offsetArr[begin], new Vertex(TAG_PEOPLE, name, ATTRIBUTE, CoreDictionary.NR_WORD_ID), wordNetAll);
});
}
