//Double check the opinion words 80%
protected string GetDoublePropagationOpinionWord(string OpinionWord, int index, string sentence)
{
FillConjunctions();
Lexicon();
if (!(index + 1 >= WordsInSentence.Length))
{
if (Conjunctions.Contains(WordsInSentence[index + 1]) && OpinionLexicon.Contains(OpinionWord))
{
return(OpinionWord);
}
else if (OpinionTargetList.Contains(WordsInSentence[index + 1]) && OpinionLexicon.Contains(OpinionWord))
{
return(OpinionWord);
}
else
{
return(null);
}
}
return(null);
}
public bool Evaluate(ReadOnlyDictionary <string, string[]> properties)
{
return(Negate ^ ((Conjunctions != null ? Conjunctions.All(c => c.Evaluate(properties)) : true) &&
(Disjunctions != null ? Disjunctions.All(d => d.Evaluate(properties)) : true) &&
(Propositions != null ? Propositions.All(p => p.Evaluate(properties)) : true)));
}
/// <summary>
/// join conjunctions to surrounding pieces, e.g.:
/// ['Mr. and Mrs.'], ['King of the Hill'], ['Jack and Jill'], ['Velasquez y Garcia']
/// </summary>
/// <param name="pieces">name pieces strings after split on spaces</param>
/// <param name="additionalPartsCount"></param>
/// <returns>new list with piece next to conjunctions merged into one piece with spaces in it.</returns>
internal static string[] joinOnConjunctions(List <string> pieces, int additionalPartsCount = 0)
{
var length = pieces.Count() + additionalPartsCount;
// don't join on conjuctions if there are only 2 parts
if (length < 3)
{
return(pieces.ToArray());
}
foreach (var conj in pieces.Where(IsConjunction).Reverse())
{
// loop through the pieces backwards, starting at the end of the list.
// Join conjunctions to the pieces on either side of them.
if (conj.Length == 1 && pieces.Count(IsRootname) < 4)
{
// if there are only 3 total parts (minus known titles, suffixes and prefixes)
// and this conjunction is a single letter, prefer treating it as an initial
// rather than a conjunction.
// http://code.google.com/p/python-nameparser/issues/detail?id=11
continue;
}
var index = pieces.IndexOf(conj);
if (index == -1)
{
continue;
}
if (index < pieces.Count() - 1)
{
// if this is not the last piece
string newPiece;
if (index == 0)
{
// if this is the first piece and it's a conjunction
var nxt = pieces[index + 1];
var cons = IsTitle(nxt) ? Conjunctions : Titles;
newPiece = string.Join(" ", pieces.Take(2));
cons.Add(newPiece);
pieces[index] = newPiece;
pieces.RemoveAt(index + 1);
continue;
}
if (IsConjunction(pieces[index - 1]))
{
// if the piece in front of this one is a conjunction too,
// add new_piece (this conjuction and the following piece)
// to the conjuctions constant so that it is recognized
// as a conjunction in the next loop.
// e.g. for ["Lord","of","the Universe"], put "the Universe"
// into the conjunctions constant.
newPiece = string.Join(" ", pieces.Skip(index).Take(2));
Conjunctions.Add(newPiece);
pieces[index] = newPiece;
pieces.RemoveAt(index + 1);
continue;
}
newPiece = string.Join(" ", pieces.Skip(index - 1).Take(3));
if (IsTitle(pieces[index - 1]))
{
// if the second name is a title, assume the first one is too and add the
// two titles with the conjunction between them to the titles constant
// so the combo we just created gets parsed as a title.
// e.g. "Mr. and Mrs." becomes a title.
Titles.Add(newPiece);
}
pieces[index - 1] = newPiece;
pieces.RemoveAt(index);
pieces.RemoveAt(index);
}
}
// join prefixes to following lastnames: ['de la Vega'], ['van Buren']
var prefixes = pieces.Where(IsPrefix).ToArray();
if (prefixes.Length > 0)
{
var i = pieces.IndexOf(prefixes[0]);
// join everything after the prefix until the next suffix
var nextSuffix = pieces.Skip(i).Where(IsSuffix).ToArray();
if (nextSuffix.Length > 0)
{
var j = pieces.IndexOf(nextSuffix[0]);
var newPiece = string.Join(" ", pieces.Skip(i).Take(j - i));
pieces = pieces
.Take(i)
.Concat(new[] { newPiece })
.Concat(pieces.Skip(j))
.ToList();
}
else
{
var newPiece = string.Join(" ", pieces.Skip(i));
pieces = pieces.Take(i).ToList();
pieces.Add(newPiece);
}
}
return(pieces.ToArray());
}
private static bool IsConjunction(string piece)
{
return(Conjunctions.Contains(piece.ToLower().Replace(".", string.Empty)) && !IsAnInitial(piece));
}
// Find an opinion word target 90%
protected string GetOpinionWordTarget(string OpinionWord, int index, string sentence)
{
FillConjunctions();
FillAdverbs();
FillComparatives();
FillDecreasers();
FillFutureWords();
FillIncreasers();
FillVerbs();
FillPronouns();
FillNegations();
OpinionWord = WordsInSentence[index];
if (OpinionWord != null)
{
if (sentence.Contains(OpinionWord))
{
int myvalue = WordsInSentence.GetUpperBound(0);
if (myvalue >= index + 1)
{
Target = WordsInSentence[index + 1];
Target = Target.ToLower();
if ((!OpinionLexicon.Contains(Target)) && (!Conjunctions.Contains(Target)) &&
(!Comparatives.Contains(Target)) &&
(!FutureWords.Contains(Target)) &&
(!Adverbs.Contains(Target)) &&
(!Increasers.Contains(Target)) &&
(!Decreasers.Contains(Target)) &&
(!Verbs.Contains(Target)) &&
(!Pronouns.Contains(Target)) &&
(!Negations.Contains(Target)))
{
return(Target);
}
{
if (Conjunctions.Contains(WordsInSentence[index + 1]))
{
Target = WordsInSentence[index + 2];
if ((!OpinionLexicon.Contains(Target)) && (!Conjunctions.Contains(Target)) &&
(!Comparatives.Contains(Target)) &&
(!FutureWords.Contains(Target)) &&
(!Adverbs.Contains(Target)) &&
(!Increasers.Contains(Target)) &&
(!Decreasers.Contains(Target)) &&
(!Verbs.Contains(Target)) &&
(!Pronouns.Contains(Target)) &&
(!Negations.Contains(Target)))
{
return(Target);
}
}
}
}
else
{
return(null);
}
}
}
return(null);
}
//Conjunction based extraction process 90%
protected List <OpinionWord> ExtractConjunctionBasedOpinionWords()
{
FillPositiveWords();
FillNegativeWords();
PositionOfReview = 0;
SentenceIndex = 0;
ConjunctionList = new List <OpinionWord>();
foreach (var Opinion in Opinions)
{
SentenceIndex = 0;
SentencesInReview = SentencesSeparator(SentencesInReview, PositionOfReview);
foreach (var sentence in SentencesInReview)
{
WordsInSentence = EachWordInSentence(WordsInSentence, SentencesInReview, SentenceIndex);
var i = -1;
foreach (var word in WordsInSentence)
{
i++;
var tempWord = word.ToLower();
var position = i;
var opinionWord = new OpinionWord(tempWord, WordOrientation)
{
OpWord = tempWord,
Orientation = WordOrientation
};
//check if the opinion word is included in filteredSeed list
var matches = FilteredSeed.Find(Didaxto => opinionWord.OpWord == tempWord);
// var matches = FilteredSeed.Where(OpinionWord => OpinionWord.OpWord == tempWord);
if (FilteredSeed.Contains(matches)) //matches.OfType<OpinionWord>().Equals(opinionWord)
{
if (position + 1 < WordsInSentence.Length)
{
if (Conjunctions.Contains(WordsInSentence[position + 1]))
{
tempWord = GetConjunctionBaseOpinionWord(tempWord, position, sentence);
{
if (PositiveWords.Contains(tempWord) || NegativeWords.Contains(tempWord))
{
WordOrientation = GetOpinionWordOrientation(tempWord, position, sentence);
if (!OpinionLexicon.Contains(tempWord))
{
ConjunctionList.Add(new OpinionWord(tempWord, WordOrientation)
{
OpWord = tempWord, Orientation = WordOrientation
});
}
else
{
ConjunctionList.Add(new OpinionWord(tempWord, opinionWord.Orientation));
}
}
}
}
}
}
}
}
SentenceIndex++;
Array.Clear(WordsInSentence, 0, WordsInSentence.Length);
}
PositionOfReview++;
return(ConjunctionList);
}
//Orientation=> if pos = true else false 80%
protected bool GetOpinionWordOrientation(string word, int position, string sentence)
{
FillPositiveWords();
FillNegativeWords();
FillAdverbs();
FillComparatives();
FillConjunctions();
FillDecreasers();
FillFutureWords();
FillIncreasers();
FillVerbs();
FillNegations();
//{pos}
if (PositiveWords.Contains(word))
{
WordOrientation = true;
}
if (NegativeWords.Contains(word))
{
WordOrientation = false;
}
if (position != 0)
{
//{pos} {pos}
if (PositiveWords.Contains(WordsInSentence[position - 1]))
{
WordOrientation = true;
}
//{neg} {pos}
if (NegativeWords.Contains(WordsInSentence[position - 1]))
{
WordOrientation = false;
}
if (position >= 2)
{
//{fut} {verb} {pos}
if (FutureWords.Contains(WordsInSentence[position - 2]) &&
Verbs.Contains(WordsInSentence[position - 1]) &&
PositiveWords.Contains(WordsInSentence[position]))
{
WordOrientation = true;
}
//{pos} {neg}
if (PositiveWords.Contains(WordsInSentence[position - 1]) &&
NegativeWords.Contains(WordsInSentence[position]))
{
WordOrientation = false;
}
//{decr} {comp} {pos}
if (Decreasers.Contains(WordsInSentence[position - 2]) && Comparatives.Contains(
WordsInSentence[position - 1]) &&
PositiveWords.Contains(
WordsInSentence[position]))
{
WordOrientation = true;
}
if (position >= 3)
{
//{pos} {conj} {nego} {cpos}
if (PositiveWords.Contains(WordsInSentence[position - 3]) && Conjunctions.Contains(
WordsInSentence[position - 2]) &&
Negations.Contains(
WordsInSentence[position - 1]) &&
PositiveWords.Contains(
WordsInSentence[position]))
{
WordOrientation = true;
}
//{neg} {conj} {incr} {cneg}
if (NegativeWords.Contains(WordsInSentence[position - 3]) && Conjunctions.Contains(
WordsInSentence[position - 2]) &&
Increasers.Contains(
WordsInSentence[position - 1]) &&
NegativeWords.Contains(
WordsInSentence[position]))
{
WordOrientation = false;
}
}
}
}
return(WordOrientation);
}
