public Sentence(IParsedReview review, SentenceItem original)
{
Review = review ?? throw new ArgumentNullException(nameof(review));
Original = original ?? throw new ArgumentNullException(nameof(original));
Index = review.Sentences.Count;
CreateNewPart();
}
public static Document Construct(this LightDocument document, IWordFactory factory)
{
var result = new Document(document.Text);
result.Author = document.Author;
result.DocumentTime = document.DocumentTime;
result.Id = document.Id;
document.Title = document.Title;
foreach (var sentence in document.Sentences)
{
var resultSentence = new SentenceItem(sentence.Text);
result.Add(resultSentence);
if (sentence.Words != null)
{
for (var i = 0; i < sentence.Words.Length; i++)
{
var word = sentence.Words[i];
var wordItem = factory.CreateWord(word.Text, word.Tag);
wordItem.WordIndex = i;
WordEx wordData = WordExFactory.Construct(wordItem);
wordData.Phrase = word.Phrase;
resultSentence.Add(wordData);
}
}
}
return(result);
}
public Document ReparseDocument(IRatingAdjustment adjustment)
{
if (adjustment?.Review?.Document == null)
{
throw new System.ArgumentNullException(nameof(adjustment));
}
adjustment.CalculateRating();
var document = new Document();
document.DocumentTime = adjustment.Review.Document.DocumentTime;
document.Stars = adjustment.Review.Document.Stars;
document.Id = adjustment.Review.Document.Id;
document.Author = adjustment.Review.Document.Author;
if (adjustment.Rating != null)
{
document.Stars = adjustment.Rating.StarsRating;
}
if (adjustment.Review.Text == null)
{
document.Text = adjustment.Review.Text;
}
foreach (var sentence in adjustment.Review.Sentences)
{
if (string.IsNullOrWhiteSpace(sentence.Text))
{
continue;
}
var sentenceItem = new SentenceItem(sentence.Text);
document.Add(sentenceItem);
foreach (var wordItem in sentence.Occurrences)
{
var word = WordExFactory.Construct(wordItem);
word.IsStop = wordItem.IsStopWord;
word.Phrase = wordItem.Parent?.Text;
word.NormalizedEntity = wordItem.NormalizedEntity;
word.Value = wordItem.Relationship.Sentiment?.DataValue?.Value;
word.IsAspect = wordItem.IsFeature;
SentimentValue value = adjustment.GetSentiment(wordItem);
if (value != null)
{
word.CalculatedValue = value.DataValue.Value;
}
else if (word.Value != null)
{
word.CalculatedValue = 0;
}
sentenceItem.Add(word);
}
}
document.Text = adjustment.Review.Document.Text;
return(document);
}
public void IsQuestion()
{
bool result = new SentenceItem("Test question").IsQuestion();
Assert.IsFalse(result);
result = new SentenceItem("Test question?").IsQuestion();
Assert.IsTrue(result);
}
public void CountCommas()
{
int result = new SentenceItem("Test question").CountCommas();
Assert.AreEqual(0, result);
result = new SentenceItem("Test, question?").CountCommas();
Assert.AreEqual(1, result);
}
public void CountSemicolons()
{
int result = new SentenceItem("Test question").CountSemicolons();
Assert.AreEqual(0, result);
result = new SentenceItem("Test, question;?").CountSemicolons();
Assert.AreEqual(1, result);
}
public void CountCharacters()
{
int result = new SentenceItem("Test question").CountCharacters();
Assert.AreEqual(12, result);
result = new SentenceItem("Test question?").CountCharacters();
Assert.AreEqual(12, result);
}
public void CountPunctuations()
{
int result = new SentenceItem("Test question").CountPunctuations();
Assert.AreEqual(0, result);
result = new SentenceItem("Test question? Test, 2, 3").CountPunctuations();
Assert.AreEqual(3, result);
}
private void CreateSentence(SentenceItem sentence)
{
if (sentence is null)
{
throw new ArgumentNullException(nameof(sentence));
}
review.AddNewSentence(sentence);
}
public void AddNewSentence(SentenceItem sentence)
{
CurrentSentence = new Sentence(this, sentence);
if (allSentences.Count > 0)
{
allSentences[allSentences.Count - 1].Next = CurrentSentence;
CurrentSentence.Previous = allSentences[allSentences.Count - 1];
}
allSentences.Add(CurrentSentence);
}
public static Document Construct(this LightDocument document, IWordFactory factory)
{
var result = new Document(document.Text);
result.Author = document.Author;
result.DocumentTime = document.DocumentTime;
result.Id = document.Id;
document.Title = document.Title;
foreach (var sentence in document.Sentences)
{
var resultSentence = new SentenceItem(sentence.Text);
result.Add(resultSentence, false);
if (sentence.Words != null)
{
for (var i = 0; i < sentence.Words.Length; i++)
{
var word = sentence.Words[i];
var wordItem = factory.CreateWord(word.Text, word.Tag);
wordItem.WordIndex = i;
WordEx wordData = WordExFactory.Construct(wordItem);
wordData.Phrase = word.Phrase;
if (!string.IsNullOrEmpty(word.Entity))
{
if (entityCache.TryGetValue(word.Entity, out var entity))
{
wordData.EntityType = entity;
}
else
{
wordData.CustomEntity = word.Entity;
}
}
else
{
wordData.EntityType = NamedEntities.None;
}
resultSentence.Add(wordData);
}
}
}
return(result);
}
private static void ProcessSentence(Document document, string sentence, IEnumerable <WordEx> words)
{
if (string.IsNullOrWhiteSpace(sentence))
{
return;
}
var currentSentence = new SentenceItem(sentence);
foreach (WordEx item in words)
{
currentSentence.Add(item);
}
if (currentSentence.Words.Count > 0)
{
document.Add(currentSentence);
}
}
public void Setup()
{
document = new Document("Test");
for (var i = 0; i < 10; i++)
{
var sentence = new SentenceItem("S1");
sentence.Add("w1");
sentence.Add("w2");
sentence.Add("w3");
document.Add(sentence, false);
foreach (var wordEx in sentence.Words)
{
wordEx.Value = 1;
}
}
info = new DirectoryInfo(Path.Combine(TestContext.CurrentContext.TestDirectory, "."));
foreach (var file in info.GetFiles("*.arff"))
{
file.Delete();
}
}
public static int CountCommas(this SentenceItem sentence)
{
return(sentence.Text.Count(item => item == ','));
}
public static int CountSemicolons(this SentenceItem sentence)
{
return(sentence.Text.Count(item => item == ';'));
}
public static int CountCharacters(this SentenceItem sentence)
{
return(sentence.Text.Count(char.IsLetterOrDigit));
}
public static int CountPunctuations(this SentenceItem sentence)
{
return(sentence.Text.Count(char.IsPunctuation));
}
/// <summary>
/// Is current sentence is question
/// </summary>
public static bool IsQuestion(this SentenceItem sentence)
{
return(sentence.Text.IndexOf('?') >= 0);
}
public Document ReparseDocument(IRatingAdjustment adjustment)
{
if (adjustment?.Review?.Document == null)
{
throw new ArgumentNullException(nameof(adjustment));
}
adjustment.CalculateRating();
var document = new Document();
document.DocumentTime = adjustment.Review.Document.DocumentTime;
document.Stars = adjustment.Review.Document.Stars;
document.Id = adjustment.Review.Document.Id;
document.Author = adjustment.Review.Document.Author;
if (adjustment.Rating != null)
{
document.Stars = adjustment.Rating.StarsRating;
}
bool buildText = false;
if (adjustment.Review.Text != null)
{
document.Text = adjustment.Review.Text;
buildText = true;
}
var vector = new SentimentVector();
foreach (var sentence in adjustment.Review.Sentences)
{
if (string.IsNullOrWhiteSpace(sentence.Text))
{
continue;
}
var sentenceItem = new SentenceItem(sentence.Text);
document.Add(sentenceItem, buildText);
foreach (var wordItem in sentence.Occurrences)
{
var word = WordExFactory.Construct(wordItem);
word.IsStop = wordItem.IsStopWord;
word.Phrase = wordItem.Parent?.Text;
word.NormalizedEntity = wordItem.NormalizedEntity;
word.Span = word.Text;
var sentiment = wordItem.Relationship.Sentiment?.DataValue;
if (sentiment != null)
{
word.Value = sentiment.Value;
word.Span = wordItem.Relationship.Sentiment.Span;
}
word.IsAspect = wordItem.IsFeature;
SentimentValue value = adjustment.GetSentiment(wordItem);
if (value != null)
{
word.CalculatedValue = value.DataValue.Value;
}
else if (word.Value != null)
{
word.CalculatedValue = 0;
}
if (adjustment.Review.Context.ExtractAttributes)
{
PopulateAttributes(vector, word, wordItem);
}
sentenceItem.Add(word);
}
}
document.Text = adjustment.Review.Document.Text;
if (adjustment.Review.Context.ExtractAttributes)
{
document.Attributes = vector.GetProbabilities().ToDictionary(item => item.Data, item => item.Probability.ToString());
}
return(document);
}
public void Setup()
{
item = new SentenceItem("Test");
}