public static void PrintSentences(SentenceSegmenter segmenter, IEnumerable<string> paragraphs)
{
foreach (var paragraph in paragraphs)
{
PrintSentences(segmenter, paragraph);
}
}
public static DetailedEvaluation Evaluate(this SentenceSegmenter segmenter, string taggedParagraph,
string tag = "#")
{
IEnumerable <int> realIndices = GetBoundaryIndices(taggedParagraph, tag);
string untaggedParagraph = taggedParagraph.Replace(tag, "");
return(Evaluate(segmenter, untaggedParagraph, realIndices));
}
public static IEnumerable <DetailedEvaluation> Evaluate(this SentenceSegmenter segmenter,
IEnumerable <string> taggedParagraphs,
string tag = "#")
{
var evaluations = new List <DetailedEvaluation>();
foreach (string taggedParagraph in taggedParagraphs)
{
evaluations.Add(Evaluate(segmenter, taggedParagraph));
}
return(evaluations);
}
public static DetailedEvaluation Evaluate(this SentenceSegmenter segmenter, string paragraph,
IEnumerable <int> realBoundaryIndices)
{
IEnumerable <int> predictedIndices = segmenter.GetBoundaryIndices(paragraph);
// ReSharper disable PossibleMultipleEnumeration
IEnumerable <int> misses = realBoundaryIndices.Except(predictedIndices);
IEnumerable <int> falseAlarms = predictedIndices.Except(realBoundaryIndices);
IEnumerable <int> hits = realBoundaryIndices.Intersect(predictedIndices);
// ReSharper restore PossibleMultipleEnumeration
int eosCandidateCount = GetEosCharCount(segmenter.EosCandidates, paragraph);
return(new DetailedEvaluation(hits, misses, falseAlarms, eosCandidateCount, paragraph));
}
public static void PrintSentences(SentenceSegmenter segmenter, string paragraph)
{
var sentences = segmenter.GetSentences(paragraph);
foreach (var sentence in sentences)
{
Console.WriteLine(sentence);
}
}
public static void EvaluateSbd(SentenceSegmenter segmenter)
{
var taggedParagraphs = File.ReadAllLines(TaggedInput);
var evaluations = segmenter.Evaluate(taggedParagraphs);
SentenceSegmenterEvaluator.GetTotalReport(evaluations, printFalseAlarms: true);
}
public SentenceIterator(SentenceSegmenter segmenter, string paragraph)
{
_paragraph = paragraph;
_segmenter = segmenter;
}