// tarPos: start from 0, not include empty space token
public static string GetTopRankStr(string inStr, HashSet <string> candidates, CSpellApi cSpellApi, bool debugFlag, int tarPos, List <TokenObj> nonSpaceTokenList)
{
// init
int rankMode = cSpellApi.GetRankMode();
double wf1 = cSpellApi.GetOrthoScoreEdDistFac();
double wf2 = cSpellApi.GetOrthoScorePhoneticFac();
double wf3 = cSpellApi.GetOrthoScoreOverlapFac();
WordWcMap wordWcMap = cSpellApi.GetWordWcMap();
string topRankStr = inStr;
int maxCandNo = cSpellApi.GetCanMaxCandNo();
Word2Vec word2VecIm = cSpellApi.GetWord2VecIm();
Word2Vec word2VecOm = cSpellApi.GetWord2VecOm();
int contextRadius = cSpellApi.GetNw1To1ContextRadius();
bool word2VecSkipWord = cSpellApi.GetWord2VecSkipWord();
double rangeFactor = cSpellApi.GetRankNwS1RankRangeFac();
double nwS1MinOScore = cSpellApi.GetRankNwS1MinOScore();
int tarSize = 1; // only for one-to-one or split, no merge here
// get the top ranked candidate
if (candidates.Count > 0)
{
// get the top rank str by scores
switch (rankMode)
{
case CSpellApi.RANK_MODE_ORTHOGRAPHIC:
topRankStr = RankByOrthographic.GetTopRankStr(inStr, candidates, wf1, wf2, wf3);
ScoreDetailByMode.PrintOrthographicScore(inStr, candidates, maxCandNo, wf1, wf2, wf3, debugFlag);
break;
case CSpellApi.RANK_MODE_FREQUENCY:
topRankStr = RankByFrequency.GetTopRankStr(candidates, wordWcMap);
ScoreDetailByMode.PrintFrequencyScore(candidates, wordWcMap, maxCandNo, debugFlag);
break;
case CSpellApi.RANK_MODE_CONTEXT:
topRankStr = RankByContext.GetTopRankStr(inStr, candidates, tarPos, tarSize, nonSpaceTokenList, word2VecIm, word2VecOm, word2VecSkipWord, contextRadius);
ScoreDetailByMode.PrintContextScore(candidates, tarPos, tarSize, nonSpaceTokenList, word2VecIm, word2VecOm, word2VecSkipWord, contextRadius, maxCandNo, debugFlag);
break;
case CSpellApi.RANK_MODE_NOISY_CHANNEL:
topRankStr = RankByNoisyChannel.GetTopRankStr(inStr, candidates, wordWcMap, wf1, wf2, wf3);
ScoreDetailByMode.PrintNoisyChannelScore(inStr, candidates, wordWcMap, maxCandNo, wf1, wf2, wf3, debugFlag);
break;
case CSpellApi.RANK_MODE_ENSEMBLE:
topRankStr = RankByEnsemble.GetTopRankStr(inStr, candidates, wordWcMap, tarPos, tarSize, nonSpaceTokenList, word2VecIm, word2VecOm, word2VecSkipWord, contextRadius, rangeFactor, wf1, wf2, wf3);
// ensemble use same basic socre as CSpell
ScoreDetailByMode.PrintCSpellScore(inStr, candidates, wordWcMap, maxCandNo, tarPos, tarSize, nonSpaceTokenList, word2VecIm, word2VecOm, word2VecSkipWord, contextRadius, wf1, wf2, wf3, debugFlag);
break;
case CSpellApi.RANK_MODE_CSPELL:
topRankStr = RankByCSpellNonWord.GetTopRankStr(inStr, candidates, wordWcMap, tarPos, tarSize, nonSpaceTokenList, word2VecIm, word2VecOm, word2VecSkipWord, contextRadius, rangeFactor, nwS1MinOScore, wf1, wf2, wf3);
ScoreDetailByMode.PrintCSpellScore(inStr, candidates, wordWcMap, maxCandNo, tarPos, tarSize, nonSpaceTokenList, word2VecIm, word2VecOm, word2VecSkipWord, contextRadius, wf1, wf2, wf3, debugFlag);
break;
}
}
return(topRankStr);
}
// this detail does not print how cSpell really fidn the top rank
// it is sorted by CSpell score
// CSpell use the cSpell score + context and frequency to find the top
public static void PrintCSpellScore(string inStr, HashSet <string> candSet, WordWcMap wordWcMap, int maxCandNo, int tarPos, int tarSize, List <TokenObj> nonSpaceTokenList, Word2Vec word2VecIm, Word2Vec word2VecOm, bool word2VecSkipWord, int contextRadius, double wf1, double wf2, double wf3, bool debugFlag)
{
if (debugFlag == true)
{
// NW 1To1
CSpellScoreNw1To1Comparator <CSpellScore> csc = new CSpellScoreNw1To1Comparator <CSpellScore>();
HashSet <CSpellScore> cScoreSet = RankByCSpellNonWord.GetCandidateScoreSet(inStr, candSet, wordWcMap, tarPos, tarSize, nonSpaceTokenList, word2VecIm, word2VecOm, word2VecSkipWord, contextRadius, wf1, wf2, wf3, debugFlag);
var list = cScoreSet.OrderBy(x => x, csc).Take(maxCandNo).Select(obj => obj.ToString()).ToList();
foreach (var item in list)
{
DebugPrint.PrintScore(item, debugFlag);
}
}
}
public static string GetTopRankStr(string inStr, HashSet <string> candidates, WordWcMap wordWcMap, int tarPos, int tarSize, List <TokenObj> nonSpaceTokenList, Word2Vec word2VecIm, Word2Vec word2VecOm, bool word2VecSkipWord, int contextRadius, double rangeFactor, double wf1, double wf2, double wf3, bool debugFlag)
{
string topRankStr = inStr;
// get the sorted list
int compareMode = CSpellScoreNw1To1Comparator <int> .COMPARE_BY_ENSEMBLE;
List <CSpellScore> candScoreList = RankByCSpellNonWord.GetCandidateScoreList(inStr, candidates, wordWcMap, tarPos, tarSize, nonSpaceTokenList, word2VecIm, word2VecOm, word2VecSkipWord, contextRadius, compareMode, wf1, wf2, wf3, debugFlag);
// Set a range for the candidates to find all possible top rank
// use the highest context and frequecny score to final rank.
if (candScoreList.Count >= 0)
{
topRankStr = candScoreList[0].GetCandStr();
}
return(topRankStr);
}
