public static bool TryGetMatchingSoundClass(this IEnumerable <SoundClass> soundClasses, SegmentPool segmentPool, Alignment <Word, ShapeNode> alignment, int seq, int col, out SoundClass soundClass)
{
ShapeNode leftNode = alignment.GetLeftNode(seq, col);
Ngram <Segment> target = alignment[seq, col].ToNgram(segmentPool);
ShapeNode rightNode = alignment.GetRightNode(seq, col);
soundClass = soundClasses.FirstOrDefault(sc => sc.Matches(leftNode, target, rightNode));
return(soundClass != null);
}
public static SoundContext ToSoundContext(this Alignment <Word, ShapeNode> alignment, SegmentPool segmentPool, int seq, int col, IEnumerable <SoundClass> soundClasses)
{
ShapeNode leftNode = alignment.GetLeftNode(seq, col);
SoundClass leftEnv;
if (leftNode == null || !soundClasses.TryGetMatchingSoundClass(segmentPool, leftNode, out leftEnv))
{
leftEnv = null;
}
Ngram <Segment> target = alignment[seq, col].ToNgram(segmentPool);
ShapeNode rightNode = alignment.GetRightNode(seq, col);
SoundClass rightEnv;
if (rightNode == null || !soundClasses.TryGetMatchingSoundClass(segmentPool, rightNode, out rightEnv))
{
rightEnv = null;
}
return(new SoundContext(leftEnv, target, rightEnv));
}
public static bool TryGetMatchingSoundClass(this IEnumerable<SoundClass> soundClasses, SegmentPool segmentPool, Alignment<Word, ShapeNode> alignment, int seq, int col, out SoundClass soundClass)
{
ShapeNode leftNode = alignment.GetLeftNode(seq, col);
Ngram<Segment> target = alignment[seq, col].ToNgram(segmentPool);
ShapeNode rightNode = alignment.GetRightNode(seq, col);
soundClass = soundClasses.FirstOrDefault(sc => sc.Matches(leftNode, target, rightNode));
return soundClass != null;
}
public void UpdatePredictedCognacy(WordPair wordPair, IWordAlignerResult alignerResult)
{
wordPair.AlignmentNotes.Clear();
int cat1Count = 0;
int cat1And2Count = 0;
int totalCount = 0;
Alignment <Word, ShapeNode> alignment = alignerResult.GetAlignments().First();
for (int column = 0; column < alignment.ColumnCount; column++)
{
ShapeNode uLeftNode = alignment.GetLeftNode(0, column);
Ngram <Segment> u = alignment[0, column].ToNgram(_segmentPool);
ShapeNode uRightNode = alignment.GetRightNode(0, column);
ShapeNode vLeftNode = alignment.GetLeftNode(1, column);
Ngram <Segment> v = alignment[1, column].ToNgram(_segmentPool);
ShapeNode vRightNode = alignment.GetRightNode(1, column);
int cat = 3;
if (u.Equals(v))
{
cat = 1;
}
else if (IgnoredMappings.IsMapped(uLeftNode, u, uRightNode, vLeftNode, v, vRightNode))
{
cat = 0;
}
else if (u.Length == 0 || v.Length == 0)
{
if (SimilarSegments.IsMapped(uLeftNode, u, uRightNode, vLeftNode, v, vRightNode))
{
cat = 1;
}
else if (IgnoreRegularInsertionDeletion && IsRegular(wordPair, alignerResult, alignment, column, v))
{
cat = 0;
}
}
else if (u[0].Type == CogFeatureSystem.VowelType && v[0].Type == CogFeatureSystem.VowelType)
{
cat = SimilarSegments.IsMapped(uLeftNode, u, uRightNode, vLeftNode, v, vRightNode) ? 1 : 2;
}
else if (u[0].Type == CogFeatureSystem.ConsonantType && v[0].Type == CogFeatureSystem.ConsonantType)
{
if (RegularConsonantEqual)
{
if (IsRegular(wordPair, alignerResult, alignment, column, v))
{
cat = 1;
}
else if (SimilarSegments.IsMapped(uLeftNode, u, uRightNode, vLeftNode, v, vRightNode))
{
cat = 2;
}
}
else
{
if (SimilarSegments.IsMapped(uLeftNode, u, uRightNode, vLeftNode, v, vRightNode))
{
cat = IsRegular(wordPair, alignerResult, alignment, column, v) ? 1 : 2;
}
}
}
if (cat > 0 && cat < 3)
{
cat1And2Count++;
if (cat == 1)
{
cat1Count++;
}
}
wordPair.AlignmentNotes.Add(cat == 0 ? "-" : cat.ToString(CultureInfo.InvariantCulture));
if (cat > 0)
{
totalCount++;
}
}
double type1Score = (double)cat1Count / totalCount;
double type1And2Score = (double)cat1And2Count / totalCount;
wordPair.PredictedCognacy = type1Score >= 0.5 && type1And2Score >= 0.75;
wordPair.PredictedCognacyScore = (type1Score * 0.75) + (type1And2Score * 0.25);
}
public void UpdateCognicity(WordPair wordPair, IWordAlignerResult alignerResult)
{
wordPair.AlignmentNotes.Clear();
int cat1Count = 0;
int cat1And2Count = 0;
int totalCount = 0;
Alignment <Word, ShapeNode> alignment = alignerResult.GetAlignments().First();
for (int column = 0; column < alignment.ColumnCount; column++)
{
ShapeNode uLeftNode = alignment.GetLeftNode(0, column);
Ngram <Segment> u = alignment[0, column].ToNgram(_segmentPool);
ShapeNode uRightNode = alignment.GetRightNode(0, column);
ShapeNode vLeftNode = alignment.GetLeftNode(1, column);
Ngram <Segment> v = alignment[1, column].ToNgram(_segmentPool);
ShapeNode vRightNode = alignment.GetRightNode(1, column);
bool regular = wordPair.VarietyPair.SoundChangeFrequencyDistribution[alignment.ToSoundContext(_segmentPool, 0, column, alignerResult.WordAligner.ContextualSoundClasses)][v] >= 3;
int cat = 3;
if (u.Equals(v))
{
cat = 1;
}
else if (_ignoredMappings.IsMapped(uLeftNode, u, uRightNode, vLeftNode, v, vRightNode))
{
cat = 0;
}
else if (u.Length == 0 || v.Length == 0)
{
if (_similarSegments.IsMapped(uLeftNode, u, uRightNode, vLeftNode, v, vRightNode) || regular)
{
cat = _ignoreRegularInsertionDeletion ? 0 : 1;
}
}
else if (u[0].Type == CogFeatureSystem.VowelType && v[0].Type == CogFeatureSystem.VowelType)
{
cat = _similarSegments.IsMapped(uLeftNode, u, uRightNode, vLeftNode, v, vRightNode) ? 1 : 2;
}
else if (u[0].Type == CogFeatureSystem.ConsonantType && v[0].Type == CogFeatureSystem.ConsonantType)
{
if (_regularConsEqual)
{
if (regular)
{
cat = 1;
}
else if (_similarSegments.IsMapped(uLeftNode, u, uRightNode, vLeftNode, v, vRightNode))
{
cat = 2;
}
}
else
{
if (_similarSegments.IsMapped(uLeftNode, u, uRightNode, vLeftNode, v, vRightNode))
{
cat = regular ? 1 : 2;
}
}
}
if (cat > 0 && cat < 3)
{
cat1And2Count++;
if (cat == 1)
{
cat1Count++;
}
}
wordPair.AlignmentNotes.Add(cat == 0 ? "-" : cat.ToString(CultureInfo.InvariantCulture));
if (cat > 0)
{
totalCount++;
}
}
double type1Score = (double)cat1Count / totalCount;
double type1And2Score = (double)cat1And2Count / totalCount;
wordPair.AreCognatePredicted = type1Score >= 0.5 && type1And2Score >= 0.75;
wordPair.CognicityScore = (type1Score * 0.75) + (type1And2Score * 0.25);
}