//the system must know the B (begin-chunk), I (in-chunk), O (out-chunk) information for computing f-score
//since such BIO information is task-dependent, tagIndex.txt is required
static void getChunkTagMap()
{
chunkTagMap.Clear();
//read the labelMap.txt for chunk tag information
StreamReader sr = new StreamReader("tagIndex.txt");
string a = sr.ReadToEnd();
a = a.Replace("\r", "");
string[] ary = a.Split(Global.lineEndAry, StringSplitOptions.RemoveEmptyEntries);
foreach (string im in ary)
{
string[] imAry = im.Split(Global.blankAry, StringSplitOptions.RemoveEmptyEntries);
int index = int.Parse(imAry[1]);
string[] tagAry = imAry[0].Split(Global.starAry, StringSplitOptions.RemoveEmptyEntries);
string tag = tagAry[tagAry.Length - 1];//the last tag is the current tag
//merge I-tag/O-tag: no need to use diversified I-tag/O-tag in computing F-score
if (tag.StartsWith("I"))
{
tag = "I";
}
if (tag.StartsWith("O"))
{
tag = "O";
}
chunkTagMap[index] = tag;
}
sr.Close();
}
//the system must know the B (begin-chunk), I (in-chunk), O (out-chunk) information for computing f-score
//since such BIO information is task-dependent, it should be explicitly coded here
static void getChunkTagMap()
{
chunkTagMap.Clear();
/*
* Noun phrase chunk task (Sun et al. COLING 2008)'s BIO information
* O 0
* I-NP 1
* B-NP 2
*/
if (Global.taskBasedChunkInfo == "np.chunk")
{
chunkTagMap["0"] = "O";
chunkTagMap["1"] = "I";
chunkTagMap["2"] = "B";
}
/*
* biomedical named entity recognition task (Sun et al. IJCAI 2009)'s BIO information
* I-RNA 0
* O 1
* B-protein 2
* B-RNA 3
* B-cell_type 4
* B-cell_line 5
* B-DNA 6
* I-protein 7
* I-DNA 8
* I-cell_type 9
* I-cell_line 10
*/
else if (Global.taskBasedChunkInfo == "bio.ner")
{
chunkTagMap["0"] = "I";
chunkTagMap["1"] = "O";
chunkTagMap["2"] = "B1";
chunkTagMap["3"] = "B2";
chunkTagMap["4"] = "B3";
chunkTagMap["5"] = "B4";
chunkTagMap["6"] = "B5";
chunkTagMap["7"] = "I";
chunkTagMap["8"] = "I";
chunkTagMap["9"] = "I";
chunkTagMap["10"] = "I";
}
/*
* Chinese word segmentation task (Sun et al. ACL 2012)'s BIO information
* B 0
* E 1
* I 2
*/
else if (Global.taskBasedChunkInfo == "wd.seg")
{
chunkTagMap["0"] = "B";
chunkTagMap["1"] = "I";
chunkTagMap["2"] = "I";
}
else
{
throw new Exception("error");
}
}