Adds sentiment evaluation methods

src/combiner/Combiner.java

@@ -30,17 +30,14 @@ public static void main(String[] args) throws JSONException, IOException {
 
         if(args != null) {
             // Collect Youtube comments and insert them to DB
-            /* YoutubeExporter.main(args);
-            Preprocessor.preprocessComments(args[4].split("=")[1], fc); */
+            YoutubeExporter.main(args);
+            Preprocessor.preprocessComments(args[4].split("=")[1], fc);
 
 
             // Collect tweets and insert them to DB
             TwitterExporter.main(args);
             Preprocessor.preprocessTweets(args[4].split("=")[1], fc, args[5].split("=")[1]);
 
-            /*fc.exportFrequencies(); //creates frequencies.txt - sorted alphabetically
-            fc.exportFrequenciesByValue(); //creates frequenciesByValue.txt - sorted by frequencies (descending order)*/
-
             if(args.length >= 5) {
                 analyticsExtractor = new AnalyticsExtractor(args[4].split("=")[1]);
                 analyticsExtractor.analyze();

src/combiner/Preprocessor.java

@@ -117,7 +117,7 @@ private static String getSubstances(String parsedTweet, String tweet) {
      * @param input The tweet's text before the processing
      * @return The parsed tweet text
      */
-    private static String preprocessTweet(String input){
+    public static String preprocessTweet(String input){
         input = prepareText(input);
         String[] tokens = tokenizer(input);
 

src/evaluation/Evaluator.java

@@ -0,0 +1,123 @@
+package evaluation;
+
+import java.io.FileNotFoundException;
+import java.io.FileReader;
+import java.io.IOException;
+import java.util.HashMap;
+
+import combiner.Preprocessor;
+import javafx.util.Pair;
+import sentiment.*;
+
+import java.util.List;
+import java.util.Scanner;
+
+/**
+ * Created by sifantid on 15/6/2017.
+ */
+public class Evaluator {
+
+    private static HashMap<String,String> annotatedData;
+    private static Analysis analyzer;
+
+    /**
+     * Keep only relevant data from annotated set (6 basic emotions and neutral sentiment)
+     */
+    static {
+        // Prepare new annotated dataset
+        annotatedData = new HashMap<>();
+        // Read the file and keep only relevant lines (referring to 6 basic emotions or neutral)
+        try {
+            Scanner input = new Scanner(new FileReader("./resources/annotatedData.txt"));
+            String emotion;
+            String tweet;
+            String[] line;
+            while (input.hasNext()) {
+                line = input.nextLine().split("\\t"); // Each tweet of the annotated dataset
+                emotion = line[line.length-1]; // Get last word - emotion
+                tweet = line[line.length-2]; // Get the related tweet
+                if(emotion.compareTo("joy") == 0 || emotion.compareTo("surprise") == 0 ||
+                        emotion.compareTo("fear") == 0 || emotion.compareTo("anger") == 0 ||
+                        emotion.compareTo("neutral") == 0 || emotion.compareTo("sadness") == 0 ||
+                        emotion.compareTo("disgust") == 0 ) {
+                    annotatedData.put(tweet,emotion);
+                }
+            }
+        } catch (FileNotFoundException e) {
+            e.printStackTrace();
+        }
+
+        try {
+            analyzer = new Analysis();
+        } catch (IOException e) {
+            e.printStackTrace();
+        }
+    }
+
+
+    public static void main(String[] args) {
+        try {
+            System.out.println("Percentage: " + analyzeAnnotated());
+        } catch (IOException e) {
+            e.printStackTrace();
+        }
+    }
+
+
+    /**
+     * Analyzes the annotated dataset
+     * @return The percentage of correct sentiment labeling (using local algorithm) for an annotated dataset
+     * @throws IOException
+     */
+    private static double analyzeAnnotated() throws IOException {
+        int correct = 0; // The number of correct tweet sentiment
+        String parsed;
+        String maxSentiment;
+        List<Pair<String, Double>> localSentiment;
+        for(String tweet : annotatedData.keySet()) { // For each tweet check if emotion is right
+            parsed = Preprocessor.preprocessTweet(tweet); // Preprocess tweet
+            localSentiment = analyzer.sentiment(parsed); // Calculate sentiment
+            // System.out.println("Tweet: " + tweet + "\n" + localSentiment);
+            maxSentiment = calculateMaxSentiment(localSentiment); // Find main sentiment based on local algorithm
+            // System.out.println("------------------------------> " + maxSentiment);
+            // System.out.println("------------------------------> Expected Sentiment: " + annotatedData.get(tweet));
+            if(maxSentiment.toLowerCase().equals(annotatedData.get(tweet))) { // Compare local sentiment to annotation
+                correct++;
+            }
+        }
+        System.out.println("CORRECT: " + correct);
+        return calculatePercentage(correct, annotatedData.size()); // Get the final percentage of correct sentiment
+    }
+
+
+    /**
+     * Calculates the main tweet sentiment given a list of sentiment scores
+     * @param localSentiment List of sentiment and score pairs
+     * @return The main sentiment of th tweet
+     */
+    private static String calculateMaxSentiment(List<Pair<String,Double>> localSentiment) {
+        double max = 0.0;
+        String mainEmotion = "neutral";
+        for(Pair<String,Double> emotion : localSentiment) {
+            if(emotion.getValue() > max) {
+                mainEmotion = emotion.getKey();
+                max = emotion.getValue();
+            }
+        }
+        return mainEmotion;
+    }
+
+    /**
+     * Calculates a percentage
+     * @param noItems The nominator
+     * @param outOf The denominator
+     * @return The percentage
+     */
+    private static double calculatePercentage(int noItems, int outOf) {
+        if(outOf != 0) {
+            return (double) noItems / outOf;
+        } else {
+            throw new IllegalArgumentException("Division by zero");
+        }
+    }
+}

src/evaluation/package-info.java

@@ -0,0 +1,5 @@
+/**
+ * Created by sifantid on 15/6/2017.
+ * Package to evaluate the sentiment analysis method of this project
+ */
+package evaluation;

src/sentiment/Analysis.java

@@ -45,6 +45,21 @@ public Analysis(String collectionName) throws IOException {
         representativeWords = sentiment.getRepresentativeWords();
     }
 
+    /**
+     * Constructor of the class; Initializes all variables
+     * @throws IOException
+     */
+    public Analysis() throws IOException {
+        emotions = Emotions.values();
+
+        String senticNetFilename = "./data/senticnet3.rdf.xml";
+        senticNetLib = new SenticNet(senticNetFilename);
+
+        SentimentAnalysis sentiment = new SentimentAnalysis();
+
+        representativeWords = sentiment.getRepresentativeWords();
+    }
+
     /**
      * Sentiment analysis for the whole case
      * @throws JSONException In case a field cannot be found
@@ -191,7 +206,7 @@ private void analyze(String dbType) throws IOException {
      * @return List of pairs (Emotion,Score) for the given tweet
      * @throws IOException
      */
-    private List<Pair<String, Double>> sentiment(String tweet) throws  IOException {
+    public List<Pair<String, Double>> sentiment(String tweet) throws  IOException {
         tweet = tweet.concat(" ");
         List<Pair<String, Double>> scores = new ArrayList<>();