private void AddWord(IWordItem occurrence, bool last)
{
if (occurrence is null)
{
throw new ArgumentNullException(nameof(occurrence));
}
if (!occurrence.IsSimple)
{
throw new ArgumentOutOfRangeException(nameof(occurrence));
}
if (!last &&
CanSplit &&
occurrence.IsConjunction())
{
review.CurrentSentence.CreateNewPart();
}
review.CurrentSentence.CurrentPart.AddItem(occurrence);
}
public static bool CanNotBeAttribute(this IWordItem word)
{
if (word.IsEmoticon())
{
return(false);
}
return
(word.IsFixed ||
word.IsQuestion ||
word.IsInvertor ||
word.IsStopWord ||
word.QuantValue.HasValue ||
word.IsItemBelonging() ||
!word.Text.HasLetters() ||
word.Text.Length < 2 || // feature is at least 3 letters
word.Entity != NamedEntities.None ||
word.POS.WordType == WordType.Symbol ||
word.IsConjunction() ||
IsNoise(word));
}
public static bool CanNotBeFeature(this IWordItem word)
{
if (word.Entity == NamedEntities.Location ||
word.Entity == NamedEntities.Organization ||
word.Entity == NamedEntities.Person ||
word.Entity == NamedEntities.Hashtag)
{
return(false);
}
// it was successfully trimmed
return(word.IsVerbLook() ||
word.IsConjunction() ||
!word.Text.HasLetters() ||
// named entities can't be features
// named entities except organizations
word.POS.WordType == WordType.Symbol ||
word.POS.WordType == WordType.Number ||
word.IsStopWord ||
word.IsInvertor ||
word.QuantValue.HasValue ||
word.IsQuestion ||
word.Entity == NamedEntities.Number ||
word.Entity == NamedEntities.Percent ||
word.Entity == NamedEntities.Money ||
word.Entity == NamedEntities.Date ||
word.Entity == NamedEntities.Duration ||
word.Entity == NamedEntities.Ordinal ||
word.Entity == NamedEntities.Time ||
word.Text.Length < 2 || // feature is at least 3 letters
word.IsItemBelonging() ||
word.Text.IsEnding("thing") || // things are too generic to be features
word.IsEmoticon() ||
word.POS.Tag == "IN" ||
word.IsNoise());
}