private int GetBigramSearchHits(Dictionary <int, string> deHyphenateTokens, CorrectionCandidate candidate, bool isUsingGoogle, out string log)
{
Correction correct = new Correction();
int hits = 0;
log = "";
string phrase = "";
if (candidate.Key - 1 >= deHyphenateTokens.Keys.Min())
{
phrase = string.Format("{0} {1}", correct.ChangeOldToNewSpell(deHyphenateTokens[candidate.Key - 1]), candidate.Candidate);
int hit = GetSearchPhraseHits(phrase, isUsingGoogle);
log += string.Format("({0},{1})", phrase, hit);
hits += hit;
}
if (deHyphenateTokens.Keys.Max() >= candidate.Key + 1)
{
phrase = string.Format("{0} {1}", candidate.Candidate, correct.ChangeOldToNewSpell(deHyphenateTokens[candidate.Key + 1]));
int hit = GetSearchPhraseHits(phrase, isUsingGoogle);
log += string.Format("({0},{1})", phrase, hit);
hits += hit;
}
candidate.Hits = hits;
return(hits);
}
public string GetGoogleSuggestionFromTrigram(Dictionary <int, string> deHyphenateTokens, int key, out int totalHits, out string log)
{
Correction corr = new Correction();
log = "";
Dictionary <int, string> suggestions = new Dictionary <int, string>();
string suggestion = deHyphenateTokens[key];
string googleSuggestion = "";
totalHits = 0;
string phrase;
if (key - 2 >= deHyphenateTokens.Keys.Min() && key - 1 >= deHyphenateTokens.Keys.Min())
{
phrase = corr.ChangeOldToNewSpell(string.Format("{0} {1} {2}", deHyphenateTokens[key - 2], deHyphenateTokens[key - 1], deHyphenateTokens[key]));
int hits = GoogleSearch(phrase, deHyphenateTokens[key], out googleSuggestion);
if (googleSuggestion != "")
{
log += string.Format("({0},'{1}',{2})", googleSuggestion, phrase, hits);
return(googleSuggestion);
}
log += string.Format("({0},'{1}',{2})", googleSuggestion, phrase, hits);
totalHits += hits;
}
if (key - 1 >= deHyphenateTokens.Keys.Min() && deHyphenateTokens.Keys.Max() >= key + 1)
{
phrase = corr.ChangeOldToNewSpell(string.Format("{0} {1} {2}", deHyphenateTokens[key - 1], deHyphenateTokens[key], deHyphenateTokens[key + 1]));
int hits = GoogleSearch(phrase, deHyphenateTokens[key], out googleSuggestion);
if (googleSuggestion != "")
{
log += string.Format("({0},'{1}',{2})", googleSuggestion, phrase, hits);
return(googleSuggestion);
}
log += string.Format("({0},'{1}',{2})", googleSuggestion, phrase, hits);
totalHits += hits;
}
if (deHyphenateTokens.Keys.Max() >= key + 1 && deHyphenateTokens.Keys.Max() >= key + 2)
{
phrase = corr.ChangeOldToNewSpell(string.Format("{0} {1} {2}", deHyphenateTokens[key], deHyphenateTokens[key + 1], deHyphenateTokens[key + 2]));
int hits = GoogleSearch(phrase, deHyphenateTokens[key], out googleSuggestion);
if (googleSuggestion != "")
{
log += string.Format("({0},'{1}',{2})", googleSuggestion, phrase, hits);
return(googleSuggestion);
}
log += string.Format("({0},'{1}',{2})", googleSuggestion, phrase, hits);
totalHits += hits;
}
return(suggestion);
}
private void btCorrectIt_Click(object sender, EventArgs e)
{
if (cbMethod.SelectedItem.ToString() == "-- Choose method --")
{
MessageBox.Show("Choose method first", "", MessageBoxButtons.OK);
return;
}
if (!File.Exists(txOCR.Text.Trim()))
{
MessageBox.Show("Browse file first", "", MessageBoxButtons.OK);
return;
}
ResetControls(false);
articleFile = txOCR.Text.Trim();
Correction correction = new Correction();
Stemmer stemmer = new Stemmer();
// DeHyphenate and clean text:
string dehyphenatedText = correction.DeHyphenate(articleFile);
rtbOCR.Text = dehyphenatedText;
// for analysis:
string dehyphenatedTextGT = "";
if (File.Exists(articleFile.Substring(0, articleFile.Length - 4) + "GT.txt"))
{
articleFileGT = articleFile.Substring(0, articleFile.Length - 4) + "GT.txt";
}
articleFileName = Path.GetFileName(articleFile);
if (!string.IsNullOrEmpty(articleFileGT))
{
dehyphenatedTextGT = correction.DeHyphenate(articleFileGT);
}
// tokenize:
deHyphenateTokens = correction.GetTokensFromText(dehyphenatedText);
Regex rgx = new Regex("[^a-zA-Z]"); //omit all non alphabet word And clean word from non alphabet:
// for analysis:
Dictionary <int, string> deHyphenateTokensGT = new Dictionary <int, string>();
if (!string.IsNullOrEmpty(articleFileGT))
{
deHyphenateTokensGT = correction.GetTokensFromText(dehyphenatedTextGT);
foreach (KeyValuePair <int, string> token in deHyphenateTokens)
{
correction.InsertOCRAndTruth(articleFileName, token.Key, rgx.Replace(token.Value, ""), rgx.Replace(deHyphenateTokensGT[token.Key], ""));
}
}
// Omit non character,single char, All Capitals word, and clean word from non alphabet:
var tmp = deHyphenateTokens.Where(p => p.Value.Length > 1).ToDictionary(p => p.Key, p => p.Value);
tmp = tmp.Where(p => p.Value.Any(Char.IsLetter)).ToDictionary(p => p.Key, p => rgx.Replace(p.Value, ""));
Dictionary <int, string> cleanTokens = tmp.Where(p => !p.Value.All(Char.IsUpper)).ToDictionary(p => p.Key, p => p.Value);
// Find Suggestion:
if (cbMethod.SelectedItem.ToString().EndsWith("Hunspell"))
{
string hunspellLog = "";
// find Suggestion using Hunspell:
foreach (KeyValuePair <int, string> err in cleanTokens)
{
string errInNewSpell = correction.ChangeOldToNewSpell(err.Value).ToLowerInvariant();
List <string> hunspellSuggestions = new List <string>();
using (SpellEngine engine = new SpellEngine())
{
LanguageConfig idConfig = new LanguageConfig();
idConfig.LanguageCode = "id";
idConfig.HunspellAffFile = "id_ID.aff";
idConfig.HunspellDictFile = "id_ID.dic";
idConfig.HunspellKey = "";
engine.AddLanguage(idConfig);
bool correct = engine["id"].Spell(errInNewSpell);
if (!correct)
{
hunspellSuggestions = engine["id"].Suggest(errInNewSpell);
if (hunspellSuggestions.Count > 0 && err.Value != correction.ChangeNewToOldSpell(hunspellSuggestions[0]))
{
deHyphenateTokens[err.Key] = "[" + correction.ChangeNewToOldSpell(hunspellSuggestions[0]) + "]";
}
// for analysis:
if (!string.IsNullOrEmpty(articleFileGT))
{
correction.UpdateFields(articleFileName, err.Key, new Dictionary <string, string> {
{ getFieldNameFromOption(), rgx.Replace(deHyphenateTokens[err.Key], "") }, { getFieldNameFromOption().Replace("Correction", "Log"), hunspellLog }
});
}
}
else
{
// for analysis:
if (!string.IsNullOrEmpty(articleFileGT))
{
correction.UpdateFields(articleFileName, err.Key, new Dictionary <string, string> {
{ getFieldNameFromOption(), err.Value }, { getFieldNameFromOption().Replace("Correction", "Log"), err.Value + " is correct" }
});
}
}
}
}
ResetControls(true);
return;
}
//check only unique word (assumption:duplicate word is correct word) :
Dictionary <int, string> checkTokens = cleanTokens;
var duplicateValues = checkTokens.GroupBy(x => x.Value).Where(x => x.Count() > 1);
List <int> duplicateKeys = new List <int>();
foreach (var item in checkTokens)
{
foreach (var dup in duplicateValues)
{
if (item.Value == dup.Key)
{
duplicateKeys.Add(item.Key);
}
}
}
foreach (var dupkey in duplicateKeys)
{
// for analysis
if (!string.IsNullOrEmpty(articleFileGT))
{
correction.UpdateFields(articleFileName, dupkey, new Dictionary <string, string> {
{ "NCorrection", checkTokens[dupkey] }, { "NLog", "Duplicate" }, { "Correction", checkTokens[dupkey] }, { "Log", "Duplicate" }, { "WOSearchCorrection", checkTokens[dupkey] }, { "WOSearchLog", "Duplicate" }, { "WOStemCorrection", checkTokens[dupkey] }, { "WOStemLog", "Duplicate" }, { "WOStemSearchCorrection", checkTokens[dupkey] }, { "WOStemSearchLog", "Duplicate" }, { "GooglePureCorrection", checkTokens[dupkey] }, { "GooglePureLog", "Duplicate" }
});
}
checkTokens.Remove(dupkey);
}
//Check Word using Dictionary(kbbi+kompas pilihan, entitas kota,negara, nama pahlawan dari wiki ):
errors = new Dictionary <int, string>();
foreach (KeyValuePair <int, string> token in checkTokens)
{
// change Soewandi to Modern Spelling:
string wordToCheck = correction.ChangeOldToNewSpell(token.Value).ToLowerInvariant();
// check word in Dictionary and Add to Error list if not there:
int frequency;
if (!correction.CheckUnigram(wordToCheck, getSQLQueryToCheckUnigram(), out frequency))
{
if (cbMethod.SelectedItem.ToString().Contains("Stemmer"))
{
// check again its stem in dictionary :
string stem = stemmer.Stemming(wordToCheck);
if (wordToCheck != stem && stemmer.checkStem(stem))
{
// for analysis
if (!string.IsNullOrEmpty(articleFileGT))
{
correction.UpdateFields(articleFileName, token.Key, new Dictionary <string, string> {
{ getFieldNameFromOption(), token.Value }, { getFieldNameFromOption().Replace("Correction", "Log"), stem + " is word" }
});
}
}
else // jika tidak ada di kamus:
{
errors.Add(token.Key, wordToCheck);
}
}
else
{
errors.Add(token.Key, wordToCheck);
}
}
else // jika ada di kamus:
{
// for analysis
if (!string.IsNullOrEmpty(articleFileGT))
{
correction.UpdateFields(articleFileName, token.Key, new Dictionary <string, string> {
{ getFieldNameFromOption(), token.Value }, { getFieldNameFromOption().Replace("Correction", "Log"), wordToCheck + " is correct" }
});
}
}
}
// Find Suggestion:
if (cbMethod.SelectedItem.ToString().EndsWith("Google"))
{
timerGoogle.Enabled = true;
indexTimerGoogle = 0;
return;
}
else
{
foreach (KeyValuePair <int, string> err in errors)
{
//get suggestion:
string log; string suggestion;
suggestion = correction.GetSuggestion(getSPNameForGetCandidates(), deHyphenateTokens, err.Key, err.Value, out log, getWithStemAndAffixCorrParamFromOption(), getWithSearchParamFromOption());
// Change suggestion back to Old Spell if any suggestions:
if (log != "No candidates")
{
suggestion = correction.ChangeNewToOldSpell(suggestion);
}
// update token dic with suggestion:
if (!suggestion.Equals(deHyphenateTokens[err.Key], StringComparison.OrdinalIgnoreCase))
{
deHyphenateTokens[err.Key] = "[" + suggestion + "]";
}
// for analysis:
if (!string.IsNullOrEmpty(articleFileGT))
{
correction.UpdateFields(articleFileName, err.Key, new Dictionary <string, string> {
{ getFieldNameFromOption(), suggestion }, { getFieldNameFromOption().Replace("Correction", "Log"), log }
});
}
}
ResetControls(true);
}
}
public List <CorrectionCandidate> GetCandidatesFromAffixCorrection(int key, string error, out string log)
{
//[prefix-1] + [prefix-2] + root + [suffix] + [possessive] + [particle]
//1.Particles: -lah, -kah, -pun, -tah.
//2.Possessives: -ku, -mu, -nya.
//3.Suffixes: -i, -an, -kan.
//4.Prefixes: meN -, beN -, peN -, teN -, di -, ke -, se -.
log = "";
List <string> Prefixs = new List <string>()
{
"", "di", "ke", "se", "ber", "bel", "be", "te", "ter", "me", "mem", "men", "meng", " menge", "meny", "pe", "per", "pem", "pen", "peng", "penge", "peny", "pel", "memper", "mempel", "menter", "member", "diper", "diter", "dipel", " diber", "keber", "keter"
};
List <string> baseSuffixes = new List <string>()
{
"i", "an", "kan"
};
List <string> possessives = new List <string>()
{
"ku", "mu", "nya"
};
List <string> particles = new List <string>()
{
"lah", "kah", "pun", "tah"
};
//List<string> akhirans = new List<string>() { "", "i", "an", "kan", "ku", "mu", "nja", "lah", "kah", "pun", "tah", "iku", "imu", "inja", "anku", "anmu", "annja", "kanku", "kanmu", "kannja", "ilah", "ikah", "ipun", "itah", "anlah", "ankah", "anpun", "antah", "kanlah", "kankah", "kanpun", "kantah", "kulah", "kukah", "kupun", "kutah", " mulah", "mukah", "mupun", "mutah", " nyalah", "nyakah", "nyapun", "nyatah", "ikulah", "ikukah", "ikupun", "ikutah", "imulah", "imukah", "imupun", "imutah", "inyalah", "inyakah", "inyapun", "inyatah", "ankulah", "ankukah", "ankupun", "ankutah", "anmulah", "anmukah", "anmupun", "anmutah", "annyalah", "annyakah", "annyapun", "annyatah", "kankulah", "kankukah", "kankupun", "kankutah", "kanmulah", "kanmukah", "kanmupun", "kanmutah", "kannjalah", "kannjakah", "kannjapun", "kannjatah" };
List <string> suffixes = new List <string>()
{
""
};
suffixes.AddRange(baseSuffixes);
suffixes.AddRange(possessives);
suffixes.AddRange(particles);
foreach (string s in baseSuffixes)
{
foreach (string po in possessives)
{
suffixes.Add(s + po); // contoh: diperbaikinya
foreach (string pa in particles)
{
suffixes.Add(s + pa); //contoh:dipelukanmulah
}
}
}
foreach (string s in baseSuffixes)
{
foreach (string pa in particles)
{
suffixes.Add(s + pa); // contoh: pertahankanlah
}
}
List <CorrectionCandidate> candidates = new List <CorrectionCandidate>();
Dictionary <string, int> dicCandidates = new Dictionary <string, int>();
Correction correct = new Correction();
string rootWord = GetRootWord(correct.ChangeOldToNewSpell(error));
if (rootWord == "" || rootWord.Length < 3)
{
return(candidates);
}
foreach (string prefix in Prefixs)
{
foreach (string suffix in suffixes)
{
Affixer affixer = new Affixer();
string candidate = correct.ChangeNewToOldSpell(affixer.Affixing(correct.ChangeNewToOldSpell(rootWord), prefix, suffix));
int levenshtein = EditDistance.LevenshteinDistance(candidate, error, 2);
if (levenshtein != -1 && levenshtein <= 2)
{
if (!dicCandidates.ContainsKey(correct.ChangeOldToNewSpell(candidate)))
{
dicCandidates.Add(correct.ChangeOldToNewSpell(candidate), levenshtein);
}
}
}
}
if (dicCandidates.Count == 0)
{
return(candidates);
}
Dictionary <string, int> dicCandidateAndFreq = GetFrequencies(dicCandidates.Keys.ToArray());
foreach (KeyValuePair <string, int> can in dicCandidates)
{
int frequency = 0;
if (dicCandidateAndFreq.ContainsKey(can.Key))
{
frequency = dicCandidateAndFreq[can.Key];
}
CorrectionCandidate corrcandidate = new CorrectionCandidate
{
Key = key,
Error = error,
Candidate = can.Key,
SameBigramAmount = -1,
Frequency = frequency,
LengthDifference = Math.Abs(can.Key.Length - correct.ChangeOldToNewSpell(error).Length),
Levensthein = can.Value
};
candidates.Add(corrcandidate);
log += can.Key + "," + can.Value + "," + frequency.ToString() + ";";
}
if (log.Length > 0)
{
log = "[" + log + "]";
}
return(candidates);
}