/// <summary>
/// 添加一条词的出现记录
/// </summary>
/// <param name="word"></param>
/// <param name="sentence"></param>
public void addWordRecord(Pair word, Sentence sentence)
{
foreach (WordInfo winfo in dc.wordinfo)
{
if (word.Word == winfo.word)
{
winfo.sum++;
winfo.appearSentences.Add(sentence);
return;
}
}
WordInfo newwinfo = new WordInfo();
newwinfo.word = word.Word;
newwinfo.sum = 1;
newwinfo.wordType = word.Flag;
newwinfo.appearSentences = new List <Sentence>();
newwinfo.appearSentences.Add(sentence);
dc.wordinfo.Add(newwinfo);
}
/// <summary>
/// 用于比较词频。降序。
/// </summary>
/// <param name="obj"></param>
/// <returns></returns>
public int CompareTo(object obj)
{
//这里是按sum降序
int res = 0;
try
{
WordInfo sObj = (WordInfo)obj;
if (this.sum > sObj.sum)
{
res = -1;
}
else if (this.sum < sObj.sum)
{
res = 1;
}
}
catch (Exception ex)
{
throw new Exception("比较异常", ex.InnerException);
}
return(res);
}
/// <summary>
/// 修正分词结果
/// </summary>
public void workResetWordCut()
{
List <Pair> tmpWordInfo;
//修正：从全文分词结果来看，将被错误分割的词拼接起来
tmpWordInfo = new List <Pair>();
List <Pair> tmpChangeWordInfo = new List <Pair>();
foreach (var f in dc.fileinfo)
{
foreach (var s in f.sentences)
{
if (s == null)
{
continue;
}
foreach (var w in s.words)
{
if (w.Word.Length > 1 && isInList(tmpWordInfo, w))
{
tmpWordInfo.Add(w);
}
}
}
}
foreach (var w in tmpWordInfo)
{
resetCuts_Link(w);
}
//修正：以出现频率较高的人名为基准，将未正确分割的含人名的词分开
List <WordInfo> tmpWord = new List <WordInfo>();
foreach (var f in dc.fileinfo)
{
foreach (var s in f.sentences)
{
if (s == null)
{
continue;
}
//mainForm.print("人名修正 - 词频统计（第" + (mainForm.dc.fileinfo.IndexOf(f) + 1) + "篇" + (f.sentneces.ToList().IndexOf(s) + 1) + "句)");
foreach (var w in s.words)
{
if (w.Word.Length > 1 && w.Flag == "nr")
{
bool haveit = false;
foreach (WordInfo winfo in tmpWord)
{
if (w.Word == winfo.word)
{
winfo.sum++;
haveit = true;
break;
}
}
if (haveit)
{
continue;
}
WordInfo newwinfo = new WordInfo();
newwinfo.word = w.Word;
newwinfo.sum = 1;
newwinfo.wordType = w.Flag;
tmpWord.Add(newwinfo);
}
}
}
}
foreach (var wi in tmpWord)
{
if (wi.sum >= 10)
{
Pair p = new Pair(wi.word, wi.wordType);
resetCuts_Cut(p);
}
}
//mainForm.print("修正分词结果完毕。");
}
