public static List <SentenceSummary> AssignGeneralQuestionAnswer(SentenceSummary question, List <SentenceSummary> summaries)
{
var answer = summaries[question.Position + 1];
var answerIsYes = answer.Phrase.Any(x => Constants.AffirmativeAnswerWords.CaseInsensitiveContains(TextAnalysisUtils.GetTokenText(x)));
var answerIsNo = answer.Phrase.Any(x => Constants.NegativeAnswerWords.CaseInsensitiveContains(TextAnalysisUtils.GetTokenText(x)));
var extras = new List <SentenceSummary>();
if (answerIsYes || (!answerIsYes && !answerIsNo))
{
var phrases = summaries
.Skip(question.Position + 1)
.TakeWhile(x => !Constants.InitialWordsToIgnore.CaseInsensitiveContains(TextAnalysisUtils.GetTokenText(x.Phrase.First())) &&
x.Answer != null);
var fullAnswer = new StringBuilder();
foreach (var summary in phrases)
{
fullAnswer.Append(summary.Phrase.SkipWhile(x =>
Constants.AffirmativeAnswerWords.CaseInsensitiveContains(TextAnalysisUtils.GetTokenText(x)) ||
TextAnalysisUtils.GetTokenText(x).CaseInsensitiveEquals(" ") ||
TextAnalysisUtils.GetTokenText(x).CaseInsensitiveEquals(",") ||
TextAnalysisUtils.GetTokenText(x).CaseInsensitiveEquals("."))
.GetPhraseText()
);
}
fullAnswer.Replace("was n't", "wasn't").Replace("do n't", "don't");
var keywords = question.KeyWords;
if (!answerIsYes && !answerIsNo)
{
keywords.Add("indecision");
}
if (fullAnswer.Length > 0 && (answerIsYes || !keywords.CaseInsensitiveContains("indecision")))
{
extras.Add(new SentenceSummary()
{
KeyWords = keywords, CombinedAnswerText = fullAnswer.ToString(), Position = question.Position
});
}
}
question.CombinedAnswerText = answer.Phrase.GetPhraseText();
return(extras);
}
public static SentenceSummary SummarizeAnswer(List <ReversedToken> reversedTokens)
{
var summary = new SentenceSummary()
{
Phrase = reversedTokens,
Names = TextAnalysisUtils.FindNames(reversedTokens),
Type = SentenceType.Answer,
KeyWords = new List <string>()
.AddIfNotNull(reversedTokens.Where(x => x.OriginalToken.PartOfSpeech.Tag == PartOfSpeech.Types.Tag.X).ToList()),
Answer = new Answer()
{
HasAffirmation = reversedTokens.Any(x => Constants.AffirmativeAnswerWords.CaseInsensitiveContains(TextAnalysisUtils.GetTokenText(x))),
Numbers = TextAnalysisUtils.FindNumbers(reversedTokens),
}
};
return(summary);
}
public static List <SentenceSummary> AssignCityAnswer(SentenceSummary question, List <SentenceSummary> summaries)
{
var answer = summaries
.Skip(question.Position + 1)
.TakeWhile(x => !Constants.InitialWordsToIgnore.CaseInsensitiveContains(TextAnalysisUtils.GetTokenText(x.Phrase.First())) &&
x.Answer != null);
var extras = new List <SentenceSummary>();
var fullAnswerText = answer.First().Phrase.GetPhraseText();
var lastCharPosition = fullAnswerText.Length - 1;
var firstComma = fullAnswerText.IndexOf(',');
var cityPart = answer.First().Phrase.TakeWhile(x => !TextAnalysisUtils.GetTokenText(x).CaseInsensitiveEquals(","));
var city = TextAnalysisUtils.FindNames(cityPart.ToList()).First();
var state = null as string;
if (firstComma != -1)
{
state = fullAnswerText.Substring(firstComma + 1, lastCharPosition - firstComma - 1);
}
if (state != null)
{
extras.Add(new SentenceSummary()
{
KeyWords = new List <string>()
{
"state"
}, CombinedAnswerText = state, Position = question.Position
});
}
question.CombinedAnswerText = city;
if (!question.KeyWords.CaseInsensitiveContains("city"))
{
question.KeyWords.Add("city");
}
return(extras);
}
public static SentenceSummary SummarizeFacts(List <ReversedToken> reverseTokens, int index, List <SentenceSummary> summaries)
{
SentenceSummary summary = null;
var phrase = reverseTokens.GetPhraseText();
if (Constants.AnyQuestionWords.CaseInsensitiveContains(TextAnalysisUtils.GetTokenText(reverseTokens.First())))
{
if (Constants.FollowUpQuestionWords.Any(x => phrase.StartsWith(x, StringComparison.InvariantCultureIgnoreCase)))
{
if (summaries.GetLastQuestion()?.GeneralQuestion != null)
{
summary = SummarizeGeneralQuestion(reverseTokens);
}
else if (summaries.GetLastQuestion()?.SpecialQuestion != null)
{
summary = SummarizeSpecialQuestion(reverseTokens);
}
}
else if (Constants.SpecialQuestionWords.Contains(reverseTokens.First().OriginalToken.Text.Content))
{
summary = SummarizeSpecialQuestion(reverseTokens);
}
else if (Constants.GeneralQuestionWords.Contains(reverseTokens.First().OriginalToken.Text.Content))
{
summary = SummarizeGeneralQuestion(reverseTokens);
}
}
else
{
summary = SummarizeAnswer(reverseTokens);
}
if (summary != null)
{
summary.Position = index;
}
return(summary);
}
public static SentenceSummary SummarizeSpecialQuestion(List <ReversedToken> reverseTokens)
{
var root = reverseTokens.FirstOrDefault(x => x.OriginalToken.DependencyEdge.Label == DependencyEdge.Types.Label.Root);
var rootRedirection = reverseTokens.FirstOrDefault(x => x.OriginalToken.DependencyEdge.Label == DependencyEdge.Types.Label.Prep && root.Relations.Any(rel => rel == x.OriginalToken));
var type = root.Relations.FirstOrDefault(x => x.PartOfSpeech.Tag == PartOfSpeech.Types.Tag.Verb || x.PartOfSpeech.Tag == PartOfSpeech.Types.Tag.Adv || x.PartOfSpeech.Tag == PartOfSpeech.Types.Tag.Det);
var subject = reverseTokens.FirstOrDefault(x => x.OriginalToken.PartOfSpeech.Tag == PartOfSpeech.Types.Tag.Pron && x.Relations.Any(rel => rel == root.OriginalToken));
if (subject == null)
{
subject = reverseTokens.FirstOrDefault(x => x.OriginalToken.PartOfSpeech.Tag == PartOfSpeech.Types.Tag.Noun && root.Relations.Any(rel => rel == x.OriginalToken));
if (subject == null)
{
subject = root;
}
}
else if (rootRedirection != null && subject.OriginalToken.PartOfSpeech.Tag == PartOfSpeech.Types.Tag.Pron)
{
subject = reverseTokens.FirstOrDefault(x => x.OriginalToken.DependencyEdge.Label == DependencyEdge.Types.Label.Pobj && rootRedirection.Relations.Any(rel => rel == x.OriginalToken));
}
var subjectSpecificationPointer = reverseTokens.FirstOrDefault(x => x.OriginalToken.DependencyEdge.Label == DependencyEdge.Types.Label.Prep && subject.Relations.Any(rel => rel == x.OriginalToken));
ReversedToken subjectSpecification = null;
if (subjectSpecificationPointer != null)
{
subjectSpecification = reverseTokens.FirstOrDefault(x => x.OriginalToken.DependencyEdge.Label == DependencyEdge.Types.Label.Pobj && subjectSpecificationPointer.Relations.Any(rel => rel == x.OriginalToken));
}
else
{
subjectSpecification = reverseTokens.FirstOrDefault(x => x.OriginalToken.DependencyEdge.Label == DependencyEdge.Types.Label.Nn && subject.Relations.Any(rel => rel == x.OriginalToken));
}
if (subjectSpecification == null)
{
subjectSpecification = reverseTokens.FirstOrDefault(x => x.OriginalToken.DependencyEdge.Label == DependencyEdge.Types.Label.Amod && subject.Relations.Any(rel => rel == x.OriginalToken));
}
var subjectOwner = reverseTokens.FirstOrDefault(x => x.OriginalToken.DependencyEdge.Label == DependencyEdge.Types.Label.Poss && subject.Relations.Any(rel => rel == x.OriginalToken));
ReversedToken subjectSpecificationOwner = null;
if (subjectSpecification != null)
{
subjectSpecificationOwner = reverseTokens.FirstOrDefault(x => x.OriginalToken.DependencyEdge.Label == DependencyEdge.Types.Label.Nn && subjectSpecification.Relations.Any(rel => rel == x.OriginalToken));
}
var ownerAlternatives = new List <ReversedToken>()
{
subjectOwner, subjectSpecificationOwner
};
var ownerCanBeEitherAlternative = false;
if (subjectOwner != null)
{
ownerAlternatives.AddRange(reverseTokens.Where(x => x.OriginalToken.DependencyEdge.Label == DependencyEdge.Types.Label.Conj && subjectOwner.Relations.Any(rel => rel == x.OriginalToken)));
if (ownerAlternatives.Any())
{
ownerCanBeEitherAlternative = reverseTokens.Any(x => x.OriginalToken.DependencyEdge.Label == DependencyEdge.Types.Label.Cc && x.OriginalToken.Text.Content.Equals("or", StringComparison.InvariantCultureIgnoreCase) && subjectOwner.Relations.Any(rel => rel == x.OriginalToken));
}
}
if (subjectSpecificationOwner != null)
{
ownerAlternatives.AddRange(reverseTokens.Where(x => x.OriginalToken.DependencyEdge.Label == DependencyEdge.Types.Label.Conj && subjectSpecificationOwner.Relations.Any(rel => rel == x.OriginalToken)));
if (ownerAlternatives.Any())
{
ownerCanBeEitherAlternative = reverseTokens.Any(x => x.OriginalToken.DependencyEdge.Label == DependencyEdge.Types.Label.Cc && x.OriginalToken.Text.Content.Equals("or", StringComparison.InvariantCultureIgnoreCase) && subjectSpecificationOwner.Relations.Any(rel => rel == x.OriginalToken));
}
}
var alternativeModifierText = ownerAlternatives.Count > 1 ? (ownerCanBeEitherAlternative ? " any" : " all") : "";
var specialQuestion = new SpecialQuestion
{
Subject = subject,
SubjectSpecification = subjectSpecification,
ResponseOwnershipRestrictionAlternatives = ownerAlternatives,
ResponseOwnershipRestrictionAnyAlternative = ownerCanBeEitherAlternative,
};
return(new SentenceSummary
{
KeyWords = new List <string>()
.AddIfNotNull(root)
.AddIfNotNull(subject)
.AddIfNotNull(subjectSpecification)
.AddIfNotNull(ownerAlternatives),
Phrase = reverseTokens,
Names = TextAnalysisUtils.FindNames(reverseTokens),
Type = SentenceType.SpecialQuestion,
SpecialQuestion = specialQuestion
});
}
public static List <SentenceSummary> AssignEmergencyContactInfoAnswer(SentenceSummary question, List <SentenceSummary> summaries)
{
var name = summaries[question.Position + 1].Names.FirstOrDefault();
var phrases = summaries
.Skip(question.Position + 1)
.TakeWhile(x => !Constants.InitialWordsToIgnore.CaseInsensitiveContains(TextAnalysisUtils.GetTokenText(x.Phrase.First())) &&
x.Answer != null);
var extras = new List <SentenceSummary>();
var relationshipPhrase = phrases.FirstOrDefault(x => Constants.RelationshipTypes.CaseInsensitiveContains(x.Phrase.Select(x => x.OriginalToken.Text.Content)));
if (relationshipPhrase != null)
{
var fullPhrase = relationshipPhrase.Phrase.GetPhraseText();
foreach (var relationship in Constants.RelationshipTypes)
{
if (fullPhrase.CaseInsensitiveContains(relationship))
{
extras.Add(new SentenceSummary()
{
KeyWords = new List <string>()
{
"relationship"
}, CombinedAnswerText = relationship, Position = question.Position
});
}
}
}
question.CombinedAnswerText = name ?? "Not disclosed";
return(extras);
}
public static List <SentenceSummary> AssignAddressAnswer(SentenceSummary question, List <SentenceSummary> summaries)
{
var answer = summaries
.Skip(question.Position + 1)
.TakeWhile(x => !Constants.InitialWordsToIgnore.CaseInsensitiveContains(TextAnalysisUtils.GetTokenText(x.Phrase.First())) &&
x.Answer != null);
var extras = new List <SentenceSummary>();
var fullAnswerText = answer.First().Phrase.GetPhraseText();
var lastCharPosition = fullAnswerText.Length - 1;
var firstComma = fullAnswerText.IndexOf(',');
var secondComma = firstComma == -1 ? -1 : fullAnswerText.IndexOf(',', firstComma);
var address = fullAnswerText.Substring(0, firstComma == -1 ? lastCharPosition : firstComma);
var city = null as string;
var state = null as string;
if (firstComma != -1)
{
city = fullAnswerText.Substring(firstComma + 1, secondComma == -1 ? lastCharPosition - firstComma : secondComma);
}
if (secondComma != -1)
{
state = fullAnswerText.Substring(secondComma + 1, lastCharPosition - secondComma);
}
if (city != null)
{
extras.Add(new SentenceSummary()
{
KeyWords = new List <string>()
{
"city"
}, CombinedAnswerText = city, Position = question.Position
});
}
if (state != null)
{
extras.Add(new SentenceSummary()
{
KeyWords = new List <string>()
{
"state"
}, CombinedAnswerText = state, Position = question.Position
});
}
question.CombinedAnswerText = address;
return(extras);
}
public static void AssignFreeformAnswer(SentenceSummary question, List <SentenceSummary> summaries)
{
var answer = summaries
.Skip(question.Position + 1)
.TakeWhile(x => !Constants.InitialWordsToIgnore.CaseInsensitiveContains(TextAnalysisUtils.GetTokenText(x.Phrase.First())) &&
x.Answer != null);
var fullAnswer = new StringBuilder();
foreach (var summary in answer)
{
fullAnswer.Append(summary.Phrase.GetPhraseText());
}
question.CombinedAnswerText = fullAnswer.ToString();
}
