/**
* <summary> The method disambiguates words with multiple possible root words in its morphological parses. If the word
* is already morphologically disambiguated, the method does not disambiguate that word. The method first check
* for multiple root words by using rootWords method. If there are multiple root words, the method select the most
* occurring root word (if its occurence wrt other root words occurence is above some threshold) for that word
* using the bestRootWord method. If root word is selected, then the case for single root word is called.</summary>
* <param name="sentence">The sentence to be disambiguated automatically.</param>
*/
protected override void AutoDisambiguateMultipleRootWords(AnnotatedSentence.AnnotatedSentence sentence)
{
FsmParseList[] fsmParses = morphologicalAnalyzer.RobustMorphologicalAnalysis(sentence);
List <FsmParse> correctParses = longestRootFirstDisambiguation.Disambiguate(fsmParses);
for (int i = 0; i < sentence.WordCount(); i++)
{
AnnotatedWord word = (AnnotatedWord)sentence.GetWord(i);
if (word.GetParse() == null)
{
SetParseAutomatically(correctParses[i], word);
}
}
}
/**
* <summary> The method checks for the MONEY entities using regular expressions. After that, if the expression is a MONEY
* expression, it also assigns the previous text, which may included numbers or some monetarial texts, MONEY tag.</summary>
* <param name="sentence">The sentence for which MONEY named entities checked.</param>
*/
protected override void AutoDetectMoney(AnnotatedSentence.AnnotatedSentence sentence)
{
for (var i = 0; i < sentence.WordCount(); i++)
{
var word = (AnnotatedWord)sentence.GetWord(i);
var wordLowercase = word.GetName().ToLower(new CultureInfo("tr"));
if (word.GetParse() != null)
{
if (Word.IsMoney(wordLowercase))
{
word.SetNamedEntityType("MONEY");
var j = i - 1;
while (j >= 0)
{
AnnotatedWord previous = (AnnotatedWord)sentence.GetWord(j);
if (previous.GetParse() != null && (previous.GetName().Equals("amerikan") ||
previous.GetParse()
.ContainsTag(MorphologicalTag.REAL) ||
previous.GetParse()
.ContainsTag(MorphologicalTag.CARDINAL) ||
previous.GetParse()
.ContainsTag(MorphologicalTag.NUMBER)))
{
previous.SetNamedEntityType("MONEY");
}
else
{
break;
}
j--;
}
}
}
}
}
/**
* <summary> The method checks for the TIME entities using regular expressions. After that, if the expression is a TIME
* expression, it also assigns the previous texts, which are numbers, TIME tag.</summary>
* <param name="sentence">The sentence for which TIME named entities checked.</param>
*/
protected override void AutoDetectTime(AnnotatedSentence.AnnotatedSentence sentence)
{
for (var i = 0; i < sentence.WordCount(); i++)
{
var word = (AnnotatedWord)sentence.GetWord(i);
var wordLowercase = word.GetName().ToLower(new CultureInfo("tr"));
if (word.GetParse() != null)
{
if (Word.IsTime(wordLowercase))
{
word.SetNamedEntityType("TIME");
if (i > 0)
{
AnnotatedWord previous = (AnnotatedWord)sentence.GetWord(i - 1);
if (previous.GetParse().ContainsTag(MorphologicalTag.CARDINAL))
{
previous.SetNamedEntityType("TIME");
}
}
}
}
}
}
public WordNodePair(AnnotatedWord annotatedWord, int no)
{
_annotatedWord = annotatedWord;
ParseNodeDrawable parent;
if (GetUniversalDependency().Equals("ADVMOD"))
{
parent = new ParseNodeDrawable(new Symbol("ADVP"));
}
else if (GetUniversalDependency().Equals("ACL"))
{
parent = new ParseNodeDrawable(new Symbol("ADJP"));
}
else
{
parent = new ParseNodeDrawable(new Symbol(annotatedWord.GetParse().GetTreePos()));
}
_node = new ParseNodeDrawable(parent, annotatedWord.ToString().Replace("\\(", "-LRB-").Replace("\\)", "-RRB-"), true, 0);
parent.AddChild(_node);
_no = no;
_doneForConnect = false;
_doneForHead = false;
}
protected List <SynSet> GetCandidateSynSets(WordNet.WordNet wordNet, FsmMorphologicalAnalyzer fsm,
AnnotatedSentence.AnnotatedSentence sentence, int index)
{
AnnotatedWord twoPrevious = null, previous = null, current, twoNext = null, next = null;
var synSets = new List <SynSet>();
current = (AnnotatedWord)sentence.GetWord(index);
if (index > 1)
{
twoPrevious = (AnnotatedWord)sentence.GetWord(index - 2);
}
if (index > 0)
{
previous = (AnnotatedWord)sentence.GetWord(index - 1);
}
if (index != sentence.WordCount() - 1)
{
next = (AnnotatedWord)sentence.GetWord(index + 1);
}
if (index < sentence.WordCount() - 2)
{
twoNext = (AnnotatedWord)sentence.GetWord(index + 2);
}
synSets = wordNet.ConstructSynSets(current.GetParse().GetWord().GetName(),
current.GetParse(), current.GetMetamorphicParse(), fsm);
if (twoPrevious?.GetParse() != null && previous?.GetParse() != null)
{
synSets.AddRange(wordNet.ConstructIdiomSynSets(twoPrevious.GetParse(), previous.GetParse(),
current.GetParse(),
twoPrevious.GetMetamorphicParse(), previous.GetMetamorphicParse(), current.GetMetamorphicParse(),
fsm));
}
if (previous?.GetParse() != null && next?.GetParse() != null)
{
synSets.AddRange(wordNet.ConstructIdiomSynSets(previous.GetParse(), current.GetParse(), next.GetParse(),
previous.GetMetamorphicParse(), current.GetMetamorphicParse(), next.GetMetamorphicParse(), fsm));
}
if (next?.GetParse() != null && twoNext?.GetParse() != null)
{
synSets.AddRange(wordNet.ConstructIdiomSynSets(current.GetParse(), next.GetParse(), twoNext.GetParse(),
current.GetMetamorphicParse(), next.GetMetamorphicParse(), twoNext.GetMetamorphicParse(), fsm));
}
if (previous?.GetParse() != null)
{
synSets.AddRange(wordNet.ConstructIdiomSynSets(previous.GetParse(), current.GetParse(),
previous.GetMetamorphicParse(), current.GetMetamorphicParse(), fsm));
}
if (next?.GetParse() != null)
{
synSets.AddRange(wordNet.ConstructIdiomSynSets(current.GetParse(), next.GetParse(),
current.GetMetamorphicParse(), next.GetMetamorphicParse(), fsm));
}
return(synSets);
}
/**
* <summary> The method checks
* 1. the previous two words and the current word; the previous, current and next word, current and the next
* two words for a three word multiword expression that occurs in the Turkish wordnet.
* 2. the previous word and current word; current word and the next word for a two word multiword expression that
* occurs in the Turkish wordnet.
* 3. the current word
* if it has only one sense. If there is only one sense for that multiword expression or word; it sets that sense.</summary>
* <param name="sentence">The sentence for which word sense disambiguation will be determined automatically.</param>
*/
protected override bool AutoLabelSingleSemantics(AnnotatedSentence.AnnotatedSentence sentence)
{
AnnotatedWord twoPrevious = null, previous = null;
AnnotatedWord twoNext = null, next = null;
for (var i = 0; i < sentence.WordCount(); i++)
{
var current = (AnnotatedWord)sentence.GetWord(i);
if (i > 1)
{
twoPrevious = (AnnotatedWord)sentence.GetWord(i - 2);
}
if (i > 0)
{
previous = (AnnotatedWord)sentence.GetWord(i - 1);
}
if (i != sentence.WordCount() - 1)
{
next = (AnnotatedWord)sentence.GetWord(i + 1);
}
if (i < sentence.WordCount() - 2)
{
twoNext = (AnnotatedWord)sentence.GetWord(i + 2);
}
if (current.GetSemantic() == null && current.GetParse() != null)
{
if (previous != null && twoPrevious != null && twoPrevious.GetParse() != null &&
previous.GetParse() != null)
{
var idioms = _turkishWordNet.ConstructIdiomSynSets(twoPrevious.GetParse(),
previous.GetParse(), current.GetParse(), twoPrevious.GetMetamorphicParse(),
previous.GetMetamorphicParse(), current.GetMetamorphicParse(), _fsm);
if (idioms.Count == 1)
{
current.SetSemantic(idioms[0].GetId());
continue;
}
}
if (previous != null && previous.GetParse() != null && next != null && next.GetParse() != null)
{
var idioms = _turkishWordNet.ConstructIdiomSynSets(previous.GetParse(),
current.GetParse(), next.GetParse(), previous.GetMetamorphicParse(),
current.GetMetamorphicParse(), next.GetMetamorphicParse(), _fsm);
if (idioms.Count == 1)
{
current.SetSemantic(idioms[0].GetId());
continue;
}
}
if (next != null && next.GetParse() != null && twoNext != null && twoNext.GetParse() != null)
{
var idioms = _turkishWordNet.ConstructIdiomSynSets(current.GetParse(),
next.GetParse(), twoNext.GetParse(), current.GetMetamorphicParse(),
next.GetMetamorphicParse(), twoNext.GetMetamorphicParse(), _fsm);
if (idioms.Count == 1)
{
current.SetSemantic(idioms[0].GetId());
continue;
}
}
if (previous != null && previous.GetParse() != null)
{
var idioms = _turkishWordNet.ConstructIdiomSynSets(previous.GetParse(),
current.GetParse(), previous.GetMetamorphicParse(), current.GetMetamorphicParse(), _fsm);
if (idioms.Count == 1)
{
current.SetSemantic(idioms[0].GetId());
continue;
}
}
if (current.GetSemantic() == null && next != null && next.GetParse() != null)
{
var idioms = _turkishWordNet.ConstructIdiomSynSets(current.GetParse(),
next.GetParse(), current.GetMetamorphicParse(), next.GetMetamorphicParse(), _fsm);
if (idioms.Count == 1)
{
current.SetSemantic(idioms[0].GetId());
continue;
}
}
var meanings = _turkishWordNet.ConstructSynSets(current.GetParse().GetWord().GetName(),
current.GetParse(), current.GetMetamorphicParse(), _fsm);
if (current.GetSemantic() == null && meanings.Count == 1)
{
current.SetSemantic(meanings[0].GetId());
}
}
}
return(true);
}
public string GetTreePos()
{
return(_annotatedWord.GetParse().GetTreePos());
}
private string FindData(string dependent, string head, bool condition1, bool condition2, AnnotatedWord dependentWord, AnnotatedWord headWord)
{
if (condition1 || condition2)
{
return("PUNCT");
}
switch (dependent)
{
case "ADVP":
if (dependentWord.GetParse().GetRootPos().Equals("VERB"))
{
return("ADVCL");
}
if (dependentWord.GetParse().GetRootPos().Equals("NOUN"))
{
return("NMOD");
}
return("ADVMOD");
case "ADJP":
switch (head)
{
case "NP":
if (dependentWord.GetParse().GetRootPos().Equals("VERB"))
{
return("ACL");
}
return("AMOD");
}
return("ADVMOD");
case "PP":
switch (head)
{
case "NP":
return("CASE");
default:
if (dependentWord.GetParse() != null && dependentWord.GetParse().GetRootPos().Equals("NOUN"))
{
return("NMOD");
}
return("ADVMOD");
}
case "DP":
return("DET");
case "NP":
switch (head)
{
case "NP":
if (dependentWord.GetParse().ContainsTag(MorphologicalTag.PROPERNOUN) && headWord.GetParse().ContainsTag(MorphologicalTag.PROPERNOUN))
{
return("FLAT");
}
if (dependentWord.GetSemantic() != null && headWord.GetSemantic() != null && dependentWord.GetSemantic().Equals(headWord.GetSemantic()))
{
return("COMPOUND");
}
return("NMOD");
case "VP":
if (dependentWord.GetSemantic() != null && headWord.GetSemantic() != null && dependentWord.GetSemantic().Equals(headWord.GetSemantic()))
{
return("COMPOUND");
}
if (dependentWord.GetParse().ContainsTag(MorphologicalTag.NOMINATIVE) || dependentWord.GetParse().ContainsTag(MorphologicalTag.ACCUSATIVE))
{
return("OBJ");
}
return("OBL");
}
return("NMOD");
case "S":
switch (head)
{
case "VP":
return("CCOMP");
default:
return("DEP");
}
case "NUM":
return("NUMMOD");
case "INTJ":
return("DISCOURSE");
case "NEG":
return("NEG");
case "CONJP":
return("CC");
default:
return("DEP");
}
}
