/* From search.c:
* Find word in index file and return parsed entry in data structure.
Input word must be exact match of string in database. */
// From the WordNet Manual (http://wordnet.princeton.edu/man/wnsearch.3WN.html)
// index_lookup() finds searchstr in the index file for pos and returns a pointer
// to the parsed entry in an Index data structure. searchstr must exactly match the
// form of the word (lower case only, hyphens and underscores in the same places) in
// the index file. NULL is returned if a match is not found.
public static Index lookup(string word, PartOfSpeech pos)
{
int j;
if (word == "")
return null;
// TDMS 14 Aug 2005 - changed to allow for numbers as well
// because the database contains searches that can start with
// numerals
//if (!char.IsLetter(word[0]))
if (!char.IsLetter(word[0]) && !char.IsNumber(word[0]))
return null;
string line = WNDB.binSearch(word, pos);
if (line == null)
return null;
Index idx = new Index();
StrTok st = new StrTok(line);
idx.Wd = st.next(); /* the word */
idx.PartOfSpeech = PartOfSpeech.of(st.next()); /* the part of speech */
idx.SenseCnt = int.Parse(st.next()); /* collins count */
int ptruse_cnt = int.Parse(st.next()); /* number of pointers types */
idx.Ptruse = new PointerType[ptruse_cnt];
for (j = 0; j < ptruse_cnt; j++)
idx.Ptruse[j] = PointerType.of(st.next());
int off_cnt = int.Parse(st.next());
idx.SynsetOffsets = new int[off_cnt];
idx.TaggedSensesCount = int.Parse(st.next());
for (j = 0; j < off_cnt; j++)
idx.SynsetOffsets[j] = int.Parse(st.next());
return idx;
}
public MyWordInfo[] FindSynonyms(ref MyWordInfo pos, bool includeMorphs)
{
pos.Word = pos.Word.ToLower();
Wnlib.Index index = Wnlib.Index.lookup(pos.Word, PartOfSpeech.of(pos.Pos));
if (index == null)
{
if (!includeMorphs)
{
return(null);
}
Wnlib.MorphStr morphs = new Wnlib.MorphStr(pos.Word, Wnlib.PartOfSpeech.of(pos.Pos));
string morph = "";
while ((morph = morphs.next()) != null)
{
index = Wnlib.Index.lookup(morph, Wnlib.PartOfSpeech.of(pos.Pos));
pos.Word = morph;
if (index != null)
{
break;
}
}
}
if (index == null)
{
return(null);
}
return(LookupCandidates(index, pos));
}
public static int GetSynsetIndex(string word, PartsOfSpeech pos)
{
word = word.ToLower();
//word=RemoveBadChars (word);
Wnlib.Index index = Wnlib.Index.lookup(word, PartOfSpeech.of(pos));
if (index == null)
{
Wnlib.MorphStr morphs = new Wnlib.MorphStr(word, Wnlib.PartOfSpeech.of(pos));
string morph = "";
while ((morph = morphs.next()) != null)
{
index = Wnlib.Index.lookup(morph, Wnlib.PartOfSpeech.of(pos));
if (index != null)
{
break;
}
}
}
if (index == null)
{
return(-1);
}
else
{
return(0);
}
}
public void beginDisambiguate()
{
Disambiguate(SParseTrees);
///////////////////////////get the text of senses ///////////////////////
for (int i = 0; i < NewParseTreeSenses.Count; i++)
{
MyWordInfo[] mwiArr = (MyWordInfo[])NewParseTreeSenses[i];
ParseTree pt;
pt = (ParseTree)NewSParseTrees[i];
AddArrStems(NewSParseTrees);
for (int j = 0; j < mwiArr.Length; j++)
{
Wnlib.PartOfSpeech p = Wnlib.PartOfSpeech.of((Wnlib.PartsOfSpeech)mwiArr[j].Pos);
try
{
//i need the stems here to get index
Wnlib.Index index = Wnlib.Index.lookup(mwiArr[j].Word.ToLower(), p);
SynSet sense = new SynSet(index, mwiArr[j].Sense, null);
NodesSenses.Add(sense.defn);
SensesNos.Add(mwiArr[j].Sense);
string s = mwiArr[j].Word.ToLower() + " : " + sense.defn;
DisambRes.Add(s);
}
catch
{
try
{
Wnlib.Index index = Wnlib.Index.lookup(Stems[j], p);
SynSet sense = new SynSet(index, mwiArr[j].Sense, null);
NodesSenses.Add(sense.defn);
SensesNos.Add(mwiArr[j].Sense);
string s = Stems[j].ToLower() + " : " + sense.defn;
DisambRes.Add(s);
}
catch
{ };
};
}
Senses = NodesSenses;
}
//////////////////////////add sense text & sense no to the nodes//////////////////////////////
AddNodesSenses(NewSParseTrees);
//////////////////////////put the output parsetrees in SparseTree again//////////////////////////////
SParseTrees = NewSParseTrees;
}
/// <summary>
/// Use if sense of word is not known
/// </summary>
/// <param name="word">word to lookup</param>
/// <returns>List of Winlib.Search</returns>
public static List <Search> GetWordNetInfo(string word)
{
List <Search> SearchList = new List <Search>();
WnLexicon.WordInfo wordi = new WnLexicon.WordInfo();
wordi.partOfSpeech = Wnlib.PartsOfSpeech.Unknown;
// for each part of speech...
Wnlib.PartsOfSpeech[] enums = (Wnlib.PartsOfSpeech[])Enum.GetValues(typeof(Wnlib.PartsOfSpeech));
wordi.senseCounts = new int[enums.Length];
for (int i = 0; i < enums.Length; i++)
{
// get a valid part of speech
Wnlib.PartsOfSpeech pos = enums[i];
if (pos == Wnlib.PartsOfSpeech.Unknown)
{
continue;
}
// get an index to a synset collection
Wnlib.Index index = Wnlib.Index.lookup(word, Wnlib.PartOfSpeech.of(pos));
// none found?
if (index == null)
{
continue;
}
//Add
SearchType ser = new SearchType(false, "OVERVIEW");
Search s = new Search(word, true, Wnlib.PartOfSpeech.of(pos), ser, 0);
SearchList.Add(s);
// does this part of speech have a higher sense count?
//wordi.senseCounts[i] = index.sense_cnt;
//if (wordi.senseCounts[i] > maxCount)
//{
// maxCount = wordi.senseCounts[i];
// wordi.partOfSpeech = pos;
//}
}
return(SearchList);
}
void doRelList(Index idx, RelList rellist)
{
int i;
bool flag;
SynSet synptr;
BitSet outsenses = new BitSet(300);
prflag = true;
for (RelList rel = rellist; rel != null; rel = rel.next)
{
flag = false;
for (i = 0; i < idx.offs.Length; i++)
if (rel.senses[i] && !outsenses[i])
{
flag = true;
synptr = new SynSet(idx.offs[i], pos, "", this, i);
synptr.strsns(i + 1);
synptr.tracePtrs(PointerType.of("HYPERPTR"), pos, 0);
synptr.frames.Clear(); // TDMS 03 Jul 2006 - frames get added in wordnet.cs after filtering
// TDMS 11 Oct 2005 - build hierarchical results
senses.Add(synptr);
outsenses[i] = true;
}
if (flag)
buf += "--------------\n";
}
for (i = 0; i < idx.offs.Length; i++)
if (!outsenses[i])
{
synptr = new SynSet(idx.offs[i], pos, "", this, i);
synptr.strsns(i + 1);
synptr.tracePtrs(PointerType.of("HYPERPTR"), pos, 0);
synptr.frames.Clear(); // TDMS 03 Jul 2006 - frames get added in wordnet.cs after filtering
// TDMS 11 Oct 2005 - build hierarchical results
senses.Add(synptr);
buf += "---------------\n";
}
}
RelList addRelatives(Index idx, int rel1, int rel2, RelList rellist)
{
/* If either of the new relatives are already in a relative group,
then add the other to the existing group (transitivity).
Otherwise create a new group and add these 2 senses to it. */
RelList rel, last = null;
for (rel = rellist; rel != null; rel = rel.next)
{
if (rel.senses[rel1] || rel.senses[rel2])
{
rel.senses[rel1] = rel.senses[rel2] = true;
/* If part of another relative group, merge the groups */
for (RelList r = rellist; r != null; r = r.next)
if (r != rel && r.senses[rel1] || r.senses[rel2])
rel.senses = rel.senses.Or(r.senses);
return rellist;
}
last = rel;
}
rel = new RelList();
rel.senses[rel1] = rel.senses[rel2] = true;
if (rellist == null)
return rel;
last.next = rel;
return rellist;
}
RelList findVerbGroups(Index idx, RelList rellist)
{
int i, j, k;
/* Read all senses */
for (i = 0; i < idx.offs.Length; i++)
{
SynSet synset = new SynSet(idx.offs[i], pos, idx.wd, this, i);
/* Look for VERBGROUP ptr(s) for this sense. If found,
create group for senses, or add to existing group. */
for (j = 0; j < synset.ptrs.Length; j++)
{
Pointer p = synset.ptrs[j];
if (p.ptp.mnemonic == "VERBGROUP")
/* Need to find sense number for ptr offset */
for (k = 0; k < idx.offs.Length; k++)
if (p.off == idx.offs[k])
{
rellist = addRelatives(idx, i, k, rellist);
break;
}
}
}
return rellist;
}
// TDMS - relatives - synonyms of verb - grouped by similarity of meaning
void relatives(Index idx)
{
RelList rellist = null;
switch (pos.name)
{
case "verb":
rellist = findVerbGroups(idx, rellist);
doRelList(idx, rellist);
break;
}
}
public SynSet(Index idx, int sens, Search sch)
: this(idx.offs[sens], idx.pos, idx.wd, sch, sens)
{
}
public void beginDisambiguate()
{
Disambiguate(SParseTrees);
///////////////////////////get the text of senses ///////////////////////
for (int i = 0; i < NewParseTreeSenses.Count; i++)
{
MyWordInfo[] mwiArr = (MyWordInfo[])NewParseTreeSenses[i];
ParseTree pt;
pt = (ParseTree)NewSParseTrees[i];
AddArrStems(NewSParseTrees);
for (int j = 0; j < mwiArr.Length; j++)
{
Wnlib.PartOfSpeech p = Wnlib.PartOfSpeech.of((Wnlib.PartsOfSpeech)mwiArr[j].Pos);
try
{
ArrayList results = new ArrayList();
Wnlib.Index index;
//i need the stems here to get index
if (mwiArr[j].Pos == PartsOfSpeech.Verb)
{
SentenceParser dummysp = new SentenceParser();
results = dummysp.GetINFOfVerb(mwiArr[j].Word.ToLower());
if (results.Count > 0)
{
index = Wnlib.Index.lookup((string)results[0], p);
}
else
{
index = Wnlib.Index.lookup(mwiArr[j].Word.ToLower(), p);
}
}
else
{
index = Wnlib.Index.lookup(mwiArr[j].Word.ToLower(), p);
}
SynSet sense = new SynSet(index, mwiArr[j].Sense, null);
NodesSenses.Add(sense.defn);
SensesNos.Add(mwiArr[j].Sense);
string s;
if (results.Count > 0)
{
s = (string)results[0] + " : " + sense.defn;
}
else
{
s = mwiArr[j].Word.ToLower() + " : " + sense.defn;
}
DisambRes.Add(s);
}
catch
{
try
{
Wnlib.Index index = Wnlib.Index.lookup(Stems[j], p);
SynSet sense = new SynSet(index, mwiArr[j].Sense, null);
NodesSenses.Add(sense.defn);
SensesNos.Add(mwiArr[j].Sense);
string s = Stems[j].ToLower() + " : " + sense.defn;
DisambRes.Add(s);
}
catch
{ };
};
}
Senses = NodesSenses;
}
//////////////////////////add sense text & sense no to the nodes//////////////////////////////
AddNodesSenses(NewSParseTrees);
//////////////////////////put the output parsetrees in SparseTree again//////////////////////////////
SParseTrees = NewSParseTrees;
}
private MyWordInfo[] LookupCandidates(Wnlib.Index index, MyWordInfo pos)
{
if (pos.Sense < 0)
{
pos.Sense = 1;
}
SynSet synset = new Wnlib.SynSet(index.offs[pos.Sense - 1], index.pos, index.wd, null, pos.Sense - 1);
ArrayList lexemes = new ArrayList();
ArrayList synIndex = new ArrayList();
foreach (Lexeme obj in synset.words)
{
lexemes.Add(obj);
synIndex.Add(index.offs[pos.Sense - 1]);
}
if (index.offs.Length > 1)
{
if (lexemes.Count <= 1)
{
for (int i = 0; i < index.offs.Length; i++)
{
synset = new Wnlib.SynSet(index.offs[i], index.pos, index.wd, null, i);
foreach (Lexeme obj in synset.words)
{
synIndex.Add(index.offs[i]);
lexemes.Add(obj);
}
}
}
else
{
synset = new Wnlib.SynSet(index.offs[0], index.pos, index.wd, null, 0);
int count = 0; //get top most frequency word senses
foreach (Lexeme obj in synset.words)
{
lexemes.Add(obj);
synIndex.Add(index.offs[0]);
++count;
if (count > 4)
{
break;
}
}
}
}
ArrayList sortedSet = new ArrayList();
Hashtable trace = new Hashtable();
int hasSem = 0;
for (int i = 0; i < lexemes.Count; i++)
{
Lexeme word = (Lexeme)lexemes[i];
word.word = word.word.ToLower();
int senIndex = (int)synIndex[i];
if (senIndex != -1 && word.wnsns > 0)
{
word.semcor = new Wnlib.SemCor(word, senIndex);
lexemes[i] = word;
++hasSem;
}
if (!trace.ContainsKey(word.word))
{
if ((word.semcor != null && word.semcor.semcor > 0) || (hasSem < 4))
{
trace[word.word] = 1;
sortedSet.Add(word);
}
}
//catch
{}
}
Lexeme[] words = (Lexeme[])sortedSet.ToArray(typeof(Lexeme));
ArrayList candidates = new ArrayList();
for (int i = 0; i < words.Length; i++)
{
string word = words[i].word.Replace("_", " ");
if (word[0] <= 'Z')
{
continue;
}
MyWordInfo newpos = new MyWordInfo(word, pos.Pos);
newpos.Sense = words[i].wnsns;
if (words[i].semcor != null)
{
newpos.Frequency = words[i].semcor.semcor;
}
else
{
newpos.Frequency = 0;
}
candidates.Add(newpos);
}
if (!trace.ContainsKey(index.wd))
{
candidates.Add(pos);
}
if (candidates.Count > 1)
{
CompareLexeme comparer = new CompareLexeme();
candidates.Sort(comparer);
}
return((MyWordInfo[])candidates.ToArray(typeof(MyWordInfo)));
}
public void ConstructMapping()
{
string concept = "";
string word = "";
int ID = -1;
string senseNo = "";
string Sense = "";
string Pos = "";
WordOlogy WO = new WordOlogy();
ArrayList wordologyArr = new ArrayList();
int conceptcounter = 0;
LoadOntology();
FileStream allConceptsFile = new FileStream(_ontologyDirectoryPath + @"\AllConcepts.txt", FileMode.Open);
StreamReader allConceptsFileReader = new StreamReader(allConceptsFile);
string _wordologyDirectoryPath =
@"..\..\..\wordology\";
BinaryFormatter bf = new BinaryFormatter();
FileStream fs = new FileStream(
_wordologyDirectoryPath + "\\wordology.txt", FileMode.Create);
int indxWatcherconceptCounter = 0;
int NoMapLexConcepts = 0;
int CannotGetSenseExeption = 0;
int AllSensesMapped = 0;
while ((concept = allConceptsFileReader.ReadLine()) != null)
{
indxWatcherconceptCounter++;
string Conceptpath = _ontologyDirectoryPath + @"\" + concept[0] + @"\" + concept;
Concept C = (Concept)Onto[concept];
Property maplexProperty = C.FullProperties["ENGLISH1"];
List <MyWordInfo> maplexsenses = new List <MyWordInfo>();
MyWordInfo mwi = new MyWordInfo();
int NoOfSensesSucceeded = 0;
if (maplexProperty != null)
{
for (int i = 0; i < maplexProperty.Fillers.Count; i++)
{
string tmp = maplexProperty.Fillers[i].ScalarFiller;
char[] charr = new char[] { '-', '_' };
string[] splt = tmp.Split(charr);
//there r fillers with no type & a-bomb masalan
if (splt.Length > 1)
{
mwi = new MyWordInfo();
for (int k = 0; k < splt.Length - 2; k++)
{
mwi.Word += splt[k] + " ";
}
mwi.Word += splt[splt.Length - 2];
if (splt[splt.Length - 1].Length == 2)
{
if (splt[splt.Length - 1][0] == 'v')
{
mwi.Pos = Wnlib.PartsOfSpeech.Verb;
}
else if (splt[splt.Length - 1][0] == 'n')
{
mwi.Pos = Wnlib.PartsOfSpeech.Noun;
}
else if (splt[splt.Length - 1][0] == 'a')
{
mwi.Pos = Wnlib.PartsOfSpeech.Adj;
}
else if (splt[splt.Length - 1][0] == 'r')
{
mwi.Pos = Wnlib.PartsOfSpeech.Adv;
}
else
{
mwi.Pos = Wnlib.PartsOfSpeech.Unknown;
}
}
else
{
mwi.Pos = Wnlib.PartsOfSpeech.Unknown;
mwi.Word += " " + splt[splt.Length - 1];
}
if (i == 0 || (maplexsenses.Count > 0 && (mwi.Word != maplexsenses[maplexsenses.Count - 1].Word || mwi.Pos != maplexsenses[maplexsenses.Count - 1].Pos)))
{
maplexsenses.Add(mwi);
}
}
//ne loop 3al ontology kolaha
}
if (maplexsenses.Count > 0)
{
MyWordInfo[] maplexArray = new MyWordInfo[maplexsenses.Count];
for (int j = 0; j < maplexsenses.Count; j++)
{
maplexArray[j] = maplexsenses[j];
}
WordSenseDisambiguator wsd = new WordSenseDisambiguator();
MyWordInfo[] res = new MyWordInfo[maplexArray.Length];
res = wsd.Disambiguate(maplexArray);
int i = 0;
foreach (MyWordInfo wi in res)
{
string tmp = maplexProperty.Fillers[i].ScalarFiller;
char[] charr = new char[] { '-', '_' };
string[] splt = tmp.Split(charr);
if (splt.Length > 1 && splt[splt.Length - 1].Length == 2)
{
WO.SenseNo = splt[splt.Length - 1];
}
else
{
// "sense doesn't have POS";
}
Wnlib.PartOfSpeech p = Wnlib.PartOfSpeech.of((Wnlib.PartsOfSpeech)wi.Pos);
try
{
Wnlib.Index index = Wnlib.Index.lookup(wi.Word.ToLower(), p);
SynSet sense = new SynSet(index, res[i].Sense, null);
WO.Sense = sense.defn;
AllSensesMapped++;
NoOfSensesSucceeded++;
try
{
WO.Pos = p.name;
}
catch
{
WO.Pos = wi.Pos.ToString();
}
ID++;
WO.Word = wi.Word;
WO.ID = ID;
WO.Concept = concept;
WO.Word = word;
}
catch
{
};
if (NoOfSensesSucceeded == 0)
{
CannotGetSenseExeption++;
}
i++;
// bf.Serialize(fs, "\n" + WO);
wordologyArr.Add(WO);
}
conceptcounter++;
}
}
else
{
NoMapLexConcepts++;
//new part
Wnlib.Index index;
Wnlib.PartOfSpeech p;
Search se;
try
{
index = Wnlib.Index.lookup(concept.ToLower(), PartOfSpeech.of(PartsOfSpeech.Noun));
if (index != null)
{
WO.Pos = "noun";
Opt[] relatedness = WordsMatching.Relatedness.GetRelatedness(PartsOfSpeech.Noun);
foreach (Opt o in relatedness)
{
for (int senseNumber = 0; senseNumber < index.sense_cnt; senseNumber++)
{
se = new Search(concept, true, PartOfSpeech.of("noun"), o.sch, senseNumber);
SynSet sense = new SynSet(index, senseNumber, se);
WO.Concept = concept;
WO.Word = concept;
WO.Sense = sense.defn;
WO.ID = ID;
ID++;
NoOfSensesSucceeded++;
AllSensesMapped++;
//bf.Serialize(fs, "\n" + WO);
wordologyArr.Add(WO);
}
}
}
}
catch
{ }
try
{
index = Wnlib.Index.lookup(concept.ToLower(), PartOfSpeech.of(PartsOfSpeech.Verb));
if (index != null)
{
WO.Pos = "verb";
Opt[] relatedness = WordsMatching.Relatedness.GetRelatedness(PartsOfSpeech.Verb);
foreach (Opt o in relatedness)
{
for (int senseNumber = 0; senseNumber < index.sense_cnt; senseNumber++)
{
se = new Search(concept, true, PartOfSpeech.of("verb"), o.sch, senseNumber);
SynSet sense = new SynSet(index, senseNumber, se);
WO.Sense = sense.defn;
WO.Concept = concept;
WO.Word = concept;
WO.ID = ID;
ID++;
NoOfSensesSucceeded++;
AllSensesMapped++;
//bf.Serialize(fs, "\n" + WO);
wordologyArr.Add(WO);
}
}
}
}
catch
{ }
try
{
index = Wnlib.Index.lookup(concept.ToLower(), PartOfSpeech.of(PartsOfSpeech.Adj));
if (index != null)
{
WO.Pos = "adj";
Opt[] relatedness = WordsMatching.Relatedness.GetRelatedness(PartsOfSpeech.Adj);
foreach (Opt o in relatedness)
{
for (int senseNumber = 0; senseNumber < index.sense_cnt; senseNumber++)
{
se = new Search(concept, true, PartOfSpeech.of("adj"), o.sch, senseNumber);
SynSet sense = new SynSet(index, senseNumber, se);
WO.Sense = sense.defn;
WO.Concept = concept;
WO.Word = concept;
WO.ID = ID;
ID++;
NoOfSensesSucceeded++;
AllSensesMapped++;
//bf.Serialize(fs, "\n" + WO);
wordologyArr.Add(WO);
}
}
}
}
catch
{ }
try
{
index = Wnlib.Index.lookup(concept.ToLower(), PartOfSpeech.of(PartsOfSpeech.Adv));
if (index != null)
{
WO.Pos = "adv";
Opt[] relatedness = WordsMatching.Relatedness.GetRelatedness(PartsOfSpeech.Noun);
foreach (Opt o in relatedness)
{
for (int senseNumber = 0; senseNumber < index.sense_cnt; senseNumber++)
{
se = new Search(concept, true, PartOfSpeech.of("adv"), o.sch, senseNumber);
SynSet sense = new SynSet(index, senseNumber, se);
WO.Sense = sense.defn;
WO.Concept = concept;
WO.Word = concept;
WO.ID = ID;
ID++;
NoOfSensesSucceeded++;
AllSensesMapped++;
//bf.Serialize(fs, "\n" + WO);
wordologyArr.Add(WO);
}
}
}
}
catch
{ }
if (NoOfSensesSucceeded != 0)
{
conceptcounter++;
}
}
}//end while
allConceptsFileReader.Close();
allConceptsFile.Close();
bf.Serialize(fs, wordologyArr);
fs.Close();
MessageBox.Show("no map-lex concepts number = " + NoMapLexConcepts.ToString());
MessageBox.Show("can't getsense pos number = " + CannotGetSenseExeption.ToString());
MessageBox.Show(conceptcounter.ToString());
}
private void MapConceptsWithMapLex(string concept, Property maplexProperty)
{
MyWordInfo mwi;
WordOlogy WO = new WordOlogy();
List <MyWordInfo> maplexsenses = new List <MyWordInfo>();
int NoOfSensesSucceeded = 0;
for (int i = 0; i < maplexProperty.Fillers.Count; i++)
{
string tmp = maplexProperty.Fillers[i].ScalarFiller;
char[] charr = new char[] { '-', '_' };
string[] splt = tmp.Split(charr);
//there r fillers with no type & a-bomb masalan
if (splt.Length > 1)
{
mwi = new MyWordInfo();
for (int k = 0; k < splt.Length - 2; k++)
{
mwi.Word += splt[k] + " ";
}
mwi.Word += splt[splt.Length - 2];
if (splt[splt.Length - 1].Length == 2)
{
if (splt[splt.Length - 1][0] == 'v')
{
mwi.Pos = Wnlib.PartsOfSpeech.Verb;
}
else if (splt[splt.Length - 1][0] == 'n')
{
mwi.Pos = Wnlib.PartsOfSpeech.Noun;
}
else if (splt[splt.Length - 1][0] == 'a')
{
mwi.Pos = Wnlib.PartsOfSpeech.Adj;
}
else if (splt[splt.Length - 1][0] == 'r')
{
mwi.Pos = Wnlib.PartsOfSpeech.Adv;
}
else
{
mwi.Pos = Wnlib.PartsOfSpeech.Unknown;
}
}
else
{
mwi.Pos = Wnlib.PartsOfSpeech.Unknown;
mwi.Word += " " + splt[splt.Length - 1];
}
if (i == 0 || (maplexsenses.Count > 0 && (mwi.Word != maplexsenses[maplexsenses.Count - 1].Word || mwi.Pos != maplexsenses[maplexsenses.Count - 1].Pos)))
{
maplexsenses.Add(mwi);
}
}
//ne loop 3al ontology kolaha
}
if (maplexsenses.Count > 0)
{
MyWordInfo[] maplexArray = new MyWordInfo[maplexsenses.Count];
for (int j = 0; j < maplexsenses.Count; j++)
{
maplexArray[j] = maplexsenses[j];
}
WordSenseDisambiguator wsd = new WordSenseDisambiguator();
MyWordInfo[] res = new MyWordInfo[maplexArray.Length];
res = wsd.Disambiguate(maplexArray);
int i = 0;
foreach (MyWordInfo wi in res)
{
string tmp = maplexProperty.Fillers[i].ScalarFiller;
char[] charr = new char[] { '-', '_' };
string[] splt = tmp.Split(charr);
if (splt.Length > 1 && splt[splt.Length - 1].Length == 2)
{
WO.SenseNo = splt[splt.Length - 1];
}
else
{
// "sense doesn't have POS";
}
Wnlib.PartOfSpeech p = Wnlib.PartOfSpeech.of((Wnlib.PartsOfSpeech)wi.Pos);
try
{
Wnlib.Index index = Wnlib.Index.lookup(wi.Word.ToLower(), p);
SynSet sense = new SynSet(index, res[i].Sense, null);
WO.Sense = sense.defn;
// AllSensesMapped++;
NoOfSensesSucceeded++;
try
{
WO.Pos = p.name;
}
catch
{
WO.Pos = wi.Pos.ToString();
}
ID++;
WO.Word = wi.Word;
WO.ID = ID;
WO.Concept = concept;
wordologyArr.Add(WO);
}
catch
{
};
if (NoOfSensesSucceeded == 0)
{
CannotGetSenseExeption++;
}
i++;
//bf.Serialize(fs, "\n" + WO);
}
conceptcounter++;
}
}
public SynSet(Index idx, int sens, Search sch)
: this(idx.SynsetOffsets[sens], idx.PartOfSpeech, idx.Wd, sch, sens)
{
}
private static MyWordInfo[] LookupCandidates(Index index, MyWordInfo pos )
{
if (pos.Sense < 0) pos.Sense=1;
SynSet synset=new Wnlib.SynSet( index.SynsetOffsets[pos.Sense - 1 ], index.PartOfSpeech , index.Wd, null , pos.Sense - 1);
ArrayList lexemes=new ArrayList() ;
ArrayList synIndex=new ArrayList() ;
foreach (Lexeme obj in synset.words)
{
lexemes.Add(obj) ;
synIndex.Add(index.SynsetOffsets[pos.Sense - 1 ]);
}
if (index.SynsetOffsets.Length > 1)
{
if (lexemes.Count <= 1)
{
for(int i=0; i < index.SynsetOffsets.Length; i++ )
{
synset=new SynSet( index.SynsetOffsets[i], index.PartOfSpeech, index.Wd, null, i );
foreach (Lexeme obj in synset.words)
{
synIndex.Add(index.SynsetOffsets[i]);
lexemes.Add(obj) ;
}
}
}
else
{
synset=new SynSet( index.SynsetOffsets[0], index.PartOfSpeech, index.Wd, null, 0 );
int count=0; //get top most frequency word senses
foreach (Lexeme obj in synset.words)
{
lexemes.Add(obj) ;
synIndex.Add(index.SynsetOffsets[0]);
++count;
if (count > 4) break;
}
}
}
ArrayList sortedSet=new ArrayList() ;
Hashtable trace=new Hashtable() ;
int hasSem=0;
for (int i = 0; i < lexemes.Count; i++)
{
Lexeme word=(Lexeme)lexemes[i];
word.word=word.word.ToLower() ;
int senIndex=(int)synIndex[i];
if (senIndex != -1 && word.wnsns > 0)
{
word.semcor=new Wnlib.SemCor(word, senIndex);
lexemes[i]=word;
++hasSem;
}
if (!trace.ContainsKey(word.word) )
{
if ((word.semcor != null && word.semcor.semcor > 0 ) || (hasSem < 4))
{
trace[word.word]=1;
sortedSet.Add(word) ;
}
}
//catch
{}
}
var words = (Lexeme[]) sortedSet.ToArray(typeof (Lexeme));
ArrayList candidates=new ArrayList();
for( int i=0; i < words.Length; i++ )
{
string word=words[i].word.Replace("_", " " );
if( word[0] <= 'Z' ) continue;
MyWordInfo newpos=new MyWordInfo(word, pos.Pos) ;
newpos.Sense=words[i].wnsns;
if (words[i].semcor != null)
newpos.Frequency=words[i].semcor.semcor;
else
newpos.Frequency=0;
candidates.Add( newpos);
}
if (!trace.ContainsKey (index.Wd))
candidates.Add(pos) ;
if (candidates.Count > 1)
{
CompareLexeme comparer=new CompareLexeme();
candidates.Sort(comparer);
}
return (MyWordInfo[])candidates.ToArray( typeof(MyWordInfo) );
}