public override void AddChunk(Chunk chunk)
{
if (chunk.Len >= maxLen)
{
maxLen = chunk.Len;
base.AddChunk(chunk);
}
}
public override void AddChunk(Chunk chunk)
{
if (chunk.AvgLen >= largestAvgLen)
{
largestAvgLen = chunk.AvgLen;
base.AddChunk(chunk);
}
}
public override void AddChunk(Chunk chunk)
{
if (chunk.SumDegree >= largesetSumDegree)
{
largesetSumDegree = chunk.SumDegree;
base.AddChunk(chunk);
}
}
public override void AddChunk(Chunk chunk)
{
if (chunk.Variance <= smallestVariance)
{
smallestVariance = chunk.Variance;
base.AddChunk(chunk);
}
}
public override Chunk Segment(Sentence sen)
{
Chunk chunk = new Chunk();
char[] chs = sen.Text;
for (int k = 0; k < 3 && !sen.IsFinish; k++)
{
int offset = sen.Offset;
int maxLen = 0;
//有了 key tree 的支持可以从头开始 max match
maxLen = dic.maxMatch(chs, offset);
chunk.Words[k] = new Word(chs, sen.StartOffset, offset, maxLen + 1);
offset += maxLen + 1;
sen.Offset = offset;
}
return chunk;
}
protected override bool IsRemove(Chunk chunk)
{
return chunk.AvgLen < largestAvgLen;
}
protected override bool IsRemove(Chunk chunk)
{
return chunk.Len < maxLen;
}
protected override bool IsRemove(Chunk chunk)
{
return chunk.SumDegree < largesetSumDegree;
}
Chunk CreateChunk(Sentence sen, char[] chs, int[] tailLen, int[] offsets, CharNode[] cns)
{
Chunk ck = new Chunk();
for (int i = 0; i < 3; i++)
{
if (offsets[i] < chs.Length)
{
ck.Words[i] = new Word(chs, sen.StartOffset, offsets[i], tailLen[i] + 1);
if (tailLen[i] == 0) //单字的要取得"字频计算出自由度"
{
CharNode cn = cns[i];
if (cn != null)
{
ck.Words[i].Degree = cn.Freq;
}
}
}
}
return ck;
}
/// <summary> /// 判断chunk是否要删除 /// </summary> /// <param name="chunk"></param> /// <returns></returns> protected abstract bool IsRemove(Chunk chunk);
public virtual void AddChunk(Chunk chunk)
{
chunks.Add(chunk);
}
protected override bool IsRemove(Chunk chunk)
{
return chunk.Variance > smallestVariance;
}
