在C＃中使用Stanford NLP lib，在尝试获取情绪（正/负）时，它始终返回-1！ 知道为什么吗？

Question

I am trying to check if a statement is positive or negative using Stanford Core NLP.

I found a few references online in Java and was able to convert/code in the missing pieces to C#.

While trying to get the sentiment score - I always get -1 as the return value.

I think it could be because I was not able to convert

 Tree tree = sentence.get(SentimentCoreAnnotations.AnnotatedTree.class);

To its .NET equivalent.

java.lang.Class treeClass = new edu.stanford.nlp.trees.TreeCoreAnnotations.TreeAnnotation().getClass();

      Tree tree = (Tree)sentence.get(treeClass);

Here is the complete code:

var jarRoot = @"D:\Core NLP Files\stanford-corenlp-full-2015-04-20\stanford-corenlp-full-2015-04-20\stanford-corenlp-3.5.2-models";

        // Text for processing
        var text = txtInp.Text;

        // Annotation pipeline configuration
        var props = new java.util.Properties();

        props.setProperty("annotators", "tokenize, ssplit, pos, lemma, ner, parse, dcoref");
        props.setProperty("sutime.binders", "0");
        props.setProperty("ner.useSUTime", "false");

        // We should change current directory, so D:\Core NLP Files\stanford-corenlp-full-2015-04-20\stanford-corenlp-full-2015-04-20 could find all the model files automatically
        var curDir = Environment.CurrentDirectory;
        Directory.SetCurrentDirectory(jarRoot);
        var pipeline = new StanfordCoreNLP(props);
        Directory.SetCurrentDirectory(curDir);

        // Annotation
        var annotation = new edu.stanford.nlp.pipeline.Annotation(text);
        pipeline.annotate(annotation);

        // Result - Pretty Print
        using (var stream = new ByteArrayOutputStream())
        {
            pipeline.prettyPrint(annotation, new PrintWriter(stream));

        //Analyze the statement as positive or negative


int mainSentiment = 0;
int longest = 0;
String[] sentimentText = { "Very Negative","Negative", "Neutral", "Positive", "Very Positive"};

NumberFormat NF = new DecimalFormat("0.0000");

//for (CoreMap sentence : document.get(CoreAnnotations.SentencesAnnotation.class)) 

var sentences = annotation.get(new CoreAnnotations.SentencesAnnotation().getClass()) as ArrayList;

   foreach(CoreMap sentence in sentences )
  {
      java.lang.Class treeClass = new edu.stanford.nlp.trees.TreeCoreAnnotations.TreeAnnotation().getClass();

      Tree tree = (Tree)sentence.get(treeClass);


                **int sentiment = RNNCoreAnnotations.getPredictedClass(tree);**

    String partText = sentence.ToString();
   label1.Text = "Sentence: '" + partText + "' is rather " + sentimentText[sentiment];

    if (partText.Length > longest)
   {
        mainSentiment = sentiment;
        longest = partText.Length;
    }   
}

if (mainSentiment == 2 || mainSentiment > 4 || mainSentiment < 0) {
label1.Text = ("Overall it was sort of neutral review");
}
else if (mainSentiment > 2) {
    label1.Text = ("Overall we are happy");
}
else {
    label1.Text = ("Bottom line. We are displeased");
}


stream.close();
        }
    }

Any ideas why I maybe getting -1 as the return value for the sentiment?

Here's the updated code:-

Tree tree = (Tree)sentence.get(typeof(edu.stanford.nlp.trees.TreeCoreAnnotations.TreeAnnotation));

                int sentiment = RNNCoreAnnotations.getPredictedClass(tree);

Value of tree - {(ROOT (S (NP (NN matrix)) (VP (VBZ is) (NP (DT a) (JJ good) (NN movie)))))}

Still getting the return value as -1 while trying to determine the sentiment.

Answer 1

与“ class”字段等效的C＃是“ typeof”运算符：

Tree tree = sentence.get(typeof(SentimentCoreAnnotations.AnnotatedTree));

Answer 2

 Tree tree =(Tree)sentence.get(typeof(SentimentCoreAnnotations.SentimentAnnotatedTree));

 int sentiment = RNNCoreAnnotations.getPredictedClass(tree);

use this as the code, i tested. It works.

Answer 3

You forgot to add the sentiment annotator to the list of annotators, that's why you still getting -1 as result.

First update the code :

Tree tree =(Tree)sentence.get(typeof(SentimentCoreAnnotations.SentimentAnnotatedTree));
int sentiment = RNNCoreAnnotations.getPredictedClass(tree);

and then you also need to change this part of code :

props.setProperty("annotators", "tokenize, ssplit, pos, lemma, ner, parse, dcoref, sentiment");

I've tested it and it works!