// Get context:
// tarPos: target word position
// tarSize: no. of tokens for target word (merge should be > 1)
// inTextList: No empty space token
// w2vIm: context must use word2Vec input matrix
// radius: number of tokens before / after the tarPos
// boolean word2VecSkipWord: skip word if the word does not have wordVec
private static List <string> GetContextForTar(int tarPos, int tarSize, List <string> nonSpaceTokenList, Word2Vec w2vIm, int radius, bool word2VecSkipWord, bool allContext)
{
// output context
List <string> outContextList = new List <string>();
// 2. find context before the tar token
int tokenNo = 0;
for (int i = tarPos - 1; i >= 0; i--)
{
string inWord = nonSpaceTokenList[i];
// check if has wordVec if word2VecSkipWord = true
if ((word2VecSkipWord == false) || (w2vIm.HasWordVec(inWord) == true))
{
tokenNo++;
if ((tokenNo <= radius) || (allContext == true))
{
outContextList.Insert(0, inWord);
}
else
{
break;
}
}
}
// 3. find context after the tar token
int endPos = tarPos + tarSize; // target could be multiwords
tokenNo = 0;
for (int i = endPos; i < nonSpaceTokenList.Count; i++)
{
string inWord = nonSpaceTokenList[i];
if ((word2VecSkipWord == false) || (w2vIm.HasWordVec(inWord) == true))
{
tokenNo++;
if ((tokenNo <= radius) || (allContext == true))
{
outContextList.Add(inWord);
}
else
{
break;
}
}
}
return(outContextList);
}
// unit test driver
public static void MainTest(string[] args)
{
//String inFile = "../data/Context/word2Vec.data";
string inFile = "../data/Context/syn1n.data";
if (args.Length == 1)
{
inFile = args[0];
}
else if (args.Length > 0)
{
Console.Error.WriteLine("Usage: java Word2Vec <inFile>");
Environment.Exit(1);
}
// test
try {
Word2Vec word2Vec = new Word2Vec(inFile);
Console.WriteLine("Dimension: " + word2Vec.GetDimension());
Console.WriteLine("Word No: " + word2Vec.GetWordNo());
Console.WriteLine("Word size in WrodVec: " + word2Vec.GetWordVecMap().Keys.Count);
Console.WriteLine("HasWordVec(man): " + word2Vec.HasWordVec("man"));
Console.WriteLine("HasWordVec(king): " + word2Vec.HasWordVec("king"));
Console.WriteLine("HasWordVec(ago): " + word2Vec.HasWordVec("ago"));
Console.WriteLine("HasWordVec(a): " + word2Vec.HasWordVec("a"));
Console.WriteLine("HasWordVec(ia): " + word2Vec.HasWordVec("ia"));
Console.WriteLine("HasWordVec(m): " + word2Vec.HasWordVec("m"));
Console.WriteLine("HasWordVec(xyxy): " + word2Vec.HasWordVec("xyxy"));
} catch (Exception e) {
Console.WriteLine(e.ToString());
Console.Write(e.StackTrace);
}
}
// These are hueristic rule for real-word one-to-one correction
// check if all one-to-one words in inTerm (candidate)
// 1. must have wordVec.
private static bool Check1To1Words(string inTerm, Word2Vec word2VecOm)
{
List <string> wordList = TermUtil.ToWordList(inTerm);
bool flag = true;
foreach (string word in wordList)
{
if (word2VecOm.HasWordVec(word) == false)
{
flag = false;
break;
}
}
return(flag);
}