protected override List<Term> GetResult(Graph graph)
{
return new IndexAnalysisMerger
{
Graph = graph,
IndexAnalysis = this
}.Execute();
}
public void Learn(Graph graph, App.Crf.SplitWord splitWord)
{
this.splitWord = splitWord;
// 亚洲人名识别
if (_isAsianName)
{
FindAsianPerson(graph);
}
// 外国人名识别
if (_isForeignName)
{
FindForeignPerson(graph);
}
}
protected override List<Term> GetResult(Graph graph)
{
var result = new LinkedList<Term>();
var length = graph.Terms.Length - 1;
for (var i = 0; i < length; i++)
{
Term term;
if ((term = graph.Terms[i]) != null)
{
result.Add(term);
while ((term = term.GetNext()) != null)
{
result.Add(term);
}
}
}
return result;
}
protected override List<Term> GetResult(Graph graph)
{
var merger = Merger.Create(() =>
{
graph.WalkPath();
var result = new List<Term>();
var length = graph.Terms.Length - 1;
for (var i = 0; i < length; i++)
{
if (graph.Terms[i] != null)
{
result.Add(graph.Terms[i]);
}
}
SetRealName(graph, result);
return result;
});
return merger.Execute();
}
/// <summary>
/// 一整句话分词,用户设置的歧异优先
/// </summary>
/// <param name="temp"></param>
private void AnalysisStr(string temp)
{
var gp = new Graph(temp);
var startOffe = 0;
if (_ambiguityForest != null)
{
var gw = new GetWord(_ambiguityForest, gp.Chars);
while ((gw.GetFrontWords()) != null)
{
if (gw.Offe > startOffe)
{
Analysis(gp, startOffe, gw.Offe);
}
var @params = gw.GetParams();
startOffe = gw.Offe;
for (var i = 0; i < @params.Length; i += 2)
{
gp.AddTerm(new Term(@params[i], startOffe, new TermNatures(new TermNature(@params[i + 1], 1))));
startOffe += @params[i].Length;
}
}
}
if (startOffe < gp.Chars.Length - 1)
{
Analysis(gp, startOffe, gp.Chars.Length);
}
var result = GetResult(gp);
_terms.AddAll(result);
}
protected abstract List<Term> GetResult(Graph graph);
/// <summary>
/// 将为标准化的词语设置到分词中
/// </summary>
/// <param name="graph"></param>
/// <param name="result"></param>
protected void SetRealName(Graph graph, List<Term> result)
{
if (!MyStaticValue.IsRealName)
{
return;
}
var str = graph.RealStr;
foreach (var term in result)
{
term.RealName = str.Substring(term.Offe, term.Offe + term.Name.Length);
}
}
private void Analysis(Graph gp, int startOffe, int endOffe)
{
var chars = gp.Chars;
for (var i = startOffe; i < endOffe; i++)
{
int start;
int end;
string str;
switch (DatDictionary.Status(chars[i]))
{
case 0:
gp.AddTerm(new Term(chars[i].ToString(), i, TermNatures.Null));
break;
case 4:
start = i;
end = 1;
while (++i < endOffe && DatDictionary.Status(chars[i]) == 4)
{
end++;
}
str = WordAlert.AlertEnglish(chars, start, end);
gp.AddTerm(new Term(str, start, TermNatures.En));
i--;
break;
case 5:
start = i;
end = 1;
while (++i < endOffe && DatDictionary.Status(chars[i]) == 5)
{
end++;
}
str = WordAlert.AlertNumber(chars, start, end);
gp.AddTerm(new Term(str, start, TermNatures.M));
i--;
break;
default:
start = i;
end = i;
var c = chars[start];
while (DatDictionary.InSystem[c] > 0)
{
end++;
if (++i >= endOffe)
break;
c = chars[i];
}
if (start == end)
{
gp.AddTerm(new Term(c.ToString(), i, TermNatures.Null));
continue;
}
_getWordsImpl.SetChars(chars, start, end);
while ((str = _getWordsImpl.AllWords()) != null)
{
gp.AddTerm(new Term(str, _getWordsImpl.Offe, _getWordsImpl.GetItem()));
}
/**
* 如果未分出词.以未知字符加入到gp中
*/
if (DatDictionary.InSystem[c] > 0 || DatDictionary.Status(c) > 3)
{
i -= 1;
}
else
{
gp.AddTerm(new Term(c.ToString(), i, TermNatures.Null));
}
break;
}
}
}
private void userDefineRecognition(Graph graph, params IWoodInterface[] forests)
{
new UserDefineRecognition(graph.Terms, forests).Recognition();
graph.RemoveLittlePath();
graph.WalkPathByScore();
}
private void FindForeignPerson(Graph graph)
{
var newWords = new ForeignPersonRecognition(graph.Terms).GetNewWords();
AddListToTerm(newWords);
}