コード例 #1
0
  public void testWordFrequency() {
    WordPredictor p = new WordPredictor();

    p.train("a big brown bear".split(" "));
    p.train("a big brown bench".split(" "));
    p.train("a big yellow banana".split(" "));

    {
      List<Prediction> pns = p.predictWord("big");
      assertEquals(1, pns.size());
      assertPrediction("brown", 100d, pns.get(0));
    }

    p.train("a big yellow duck".split(" "));

    {
      List<Prediction> pns = p.predictWord("big");
      assertEquals(2, pns.size());
      assertPrediction("brown", 50d, pns.get(0));
      assertPrediction("yellow", 50d, pns.get(1));
    }

    p.train("a big yellow daisy".split(" "));

    {
      List<Prediction> pns = p.predictWord("big");
      assertEquals(2, pns.size());
      assertPrediction("yellow", (3d / 5) * 100, pns.get(0));
      assertPrediction("brown", (2d / 5) * 100, pns.get(1));
    }
  }
コード例 #2
0
  public void testWordPredictor() {
    WordPredictor p = new WordPredictor();

    {
      // Should be empty before training
      assertEquals(Collections.emptyList(), p.predictWord("a"));
    }

    p.train("a big brown bear".split(" "));
    p.train("a big brown bench".split(" "));

    {
      List<Prediction> pns = p.predictWord("a");
      assertEquals(1, pns.size());
      assertPrediction("big", 100d, pns.get(0));
    }
    {
      List<Prediction> pns = p.predictWord("big");
      assertEquals(1, pns.size());
      assertPrediction("brown", 100d, pns.get(0));
    }
    {
      // Word must occur more than once to be considered interesting.
      assertEquals(Collections.emptyList(), p.predictWord("brown"));
      assertEquals(Collections.emptyList(), p.predictWord("bear"));
      assertEquals(Collections.emptyList(), p.predictWord("foo"));
    }
  }
コード例 #3
0
  public void testReset() {
    WordPredictor p = new WordPredictor();

    p.train("a big brown bath".split(" "));
    p.train("a big brown bath".split(" "));
    {
      List<Prediction> pns = p.predictWord("a");
      assertEquals(1, pns.size());
      assertPrediction("big", 100d, pns.get(0));
    }

    p.reset();

    {
      assertEquals(Collections.emptyList(), p.predictWord("a"));
    }
  }
コード例 #4
0
  public void testEmptyInput() {
    WordPredictor p = new WordPredictor();

    {
      try {
        p.train(null);
        fail("should throw NPE on null input");
      } catch (NullPointerException e) {
        // OK
      }
    }

    p.train("a big brown bear".split(" "));
    p.train("a big brown bench".split(" "));

    {
      // Empty string should give empty predictions even after training
      assertEquals(Collections.emptyList(), p.predictWord(""));
    }

    {
      try {
        p.predictWord(null);
        fail("Should throw NPE on null input");
      } catch (NullPointerException e) {
        // OK
      }
    }
  }
  // Test main method
  public static void main(String[] args) {
    // Train a model on the first bit of Moby-Dick
    WordPredictor wp = new WordPredictor();
    System.out.println("bad1 = " + wp.getBest("the"));
    wp.train("moby_start.txt");
    System.out.println("training words = " + wp.getTrainingCount());

    // Try and crash things on bad input
    System.out.println("bad2 = " + wp.getBest("the"));
    wp.train("thisfiledoesnotexist.txt");
    System.out.println("training words = " + wp.getTrainingCount() + "\n");

    String[] words = {"the", "me", "zebra", "ishmael", "savage"};
    for (String s : words) System.out.println("count, " + s + " = " + wp.getWordCount(s));
    System.out.println();

    wp.train("moby_end.txt");

    // Check the counts again after training on the end of the book
    for (String s : words) System.out.println("count, " + s + " = " + wp.getWordCount(s));
    System.out.println();

    // Get the object ready to start looking things up
    wp.build();

    // Do some prefix lookups
    String[] test = {"a", "ab", "b", "be", "t", "th", "archang"};
    for (String prefix : test)
      System.out.println(prefix + " -> " + wp.getBest(prefix).getList().get(0).getWord());
    System.out.println("training words = " + wp.getTrainingCount() + "\n");

    // Add two individual words to the training data
    wp.trainWord("beefeater");
    wp.trainWord("BEEFEATER!");
    wp.trainWord("Pneumonoultramicroscopicsilicovolcanoconiosis");

    // The change should have no effect for prefix lookup until we build()
    System.out.println("before, b -> " + wp.getBest("b").getList().get(0).getWord());
    System.out.println("before, pn -> " + wp.getBest("pn"));
    wp.build();
    System.out.println("after, b -> " + wp.getBest("b").getList().get(0).getWord());
    System.out.println("after, pn -> " + wp.getBest("pn").getList().get(0).getWord());
    System.out.println("training words = " + wp.getTrainingCount() + "\n");

    // Test out training on a big file, timing the training as well
    Stats stats1 = new Stats();
    wp.train("mobydick.txt");
    wp.build();
    for (String prefix : test)
      System.out.println(prefix + " -> " + wp.getBest(prefix).getList().get(0).getWord());
    System.out.println("training words = " + wp.getTrainingCount());
    System.out.println(stats1);

    // Test lookup using random prefixes between 1-6 characters
    System.out.println("\nRandom load test:");
    Stats stats2 = new Stats();
    final String VALID = "abcdefghijklmnopqrstuvwxyz'";
    final long TEST_NUM = 10000000;
    long hits = 0;
    for (long i = 0; i < TEST_NUM; i++) {
      String prefix = "";
      for (int j = 0; j <= (int) (Math.random() * 6); j++)
        prefix += VALID.charAt((int) (Math.random() * VALID.length()));
      // Word word = wp.getBest(prefix).getList().get(0);
      // if (word != null)
      //	hits++;
    }
    System.out.println(stats2);
    System.out.println("Hit % = " + ((double) hits / TEST_NUM * 100.0));
  }
コード例 #6
0
  public void testMinFrequency() {
    WordPredictor p = new WordPredictor();

    for (int i = 0; i < 2; i++) {
      p.train("a big brown bear".split(" "));
      p.train("a big brown bench".split(" "));
      p.train("a big brown bazooka".split(" "));
      p.train("a big brown bazinga".split(" "));
      p.train("a big brown balloon".split(" "));
      p.train("a big brown boulder".split(" "));
      p.train("a big brown blanket".split(" "));
      p.train("a big brown balcony".split(" "));
      p.train("a big brown binder".split(" "));
      p.train("a big brown book".split(" "));
    }

    {
      List<Prediction> pns = p.predictWord("brown");
      assertEquals(10, pns.size());
      assertPrediction("balcony", 10d, pns.get(0));
      assertPrediction("boulder", 10d, pns.get(9));
    }

    // Add an eleventh distinct word after "brown", making none of the
    // predictions higher than the minimum threshold for being considered
    // interesting.
    p.train("a big brown bath".split(" "));
    p.train("a big brown bath".split(" "));

    {
      assertEquals(Collections.emptyList(), p.predictWord("brown"));
    }
    {
      List<Prediction> pns = p.predictWord("a");
      assertEquals(1, pns.size());
      assertPrediction("big", 100d, pns.get(0));
    }
  }
  public static void main(String[] args) {
    // GUI Variables
    int width = 1024;
    int height = 256;
    int pauseTime = 0;
    Font typeFont = new Font("Consolas", 1, 24);
    Font sentenceFont = new Font("Consolas", 0, 16);
    boolean multipleWords = false;
    // Strings for typing word, prediction, and sentence.
    StringBuilder typeString = new StringBuilder();
    String prediction;
    StringBuilder sentence = new StringBuilder();

    // Set up window
    StdDraw.setCanvasSize(width, height);
    StdDraw.setXscale(0.0, (double) width);
    StdDraw.setYscale(0.0, (double) height);

    // Train on files.
    WordPredictor predictor;
    if (args.length > 0) {

      try {
        int number = Integer.parseInt(args[args.length - 1]);
        if (number > 0 && number <= 9) {
          predictor = new WordPredictor(number);
          multipleWords = true;
        } else {
          predictor = new WordPredictor();
        }
      } catch (NumberFormatException e) {
        predictor = new WordPredictor();
        predictor.train(args[args.length - 1]);
      }

      for (int i = 0; i < args.length - 1; i++) {
        predictor.train(args[i]);
      }

      predictor.build();
    } else {
      predictor = new WordPredictor();
    }

    // Main loop for GUI and typing.
    while (true) {
      // Clear screen and reset variables.
      StdDraw.clear();
      ArrayList<Word> best = null;
      if (predictor.getBest(typeString.toString()) != null) {
        best = predictor.getBest(typeString.toString()).getList();
      }
      // If next key has been typed then act.
      if (StdDraw.hasNextKeyTyped()) {
        char key = StdDraw.nextKeyTyped();
        // If key is a-z or apostrophe add it to string.
        if ((key >= 'a' && key <= 'z') || key == '\'') {
          typeString.append(key);
        } else if (key == ' ') {
          // If space move word to sentence, train into dict and reset
          if (typeString.length() > 0) {
            sentence.append(typeString + " ");
            predictor.trainWord(typeString.toString());
            predictor.build();
            typeString = null;
            typeString = new StringBuilder();
          }
        } else if (key == '\b') {
          // Remove last letter if possible.
          if (typeString.length() > 0) {
            typeString.deleteCharAt(typeString.length() - 1);
          }
        } else if (key == '\n') {

          if (best != null) {
            // If enter was pressed and there is a prediction
            // Use the prediction and reset current word.
            sentence.append(best.get(0).getWord() + " ");
            typeString = null;
            typeString = new StringBuilder();
            predictor.trainWord(best.get(0).getWord());
            predictor.build();
          }
        } else if (multipleWords) {
          // If there are multiple words available listen for numbers
          // These will allow for adding from suggestions.
          if (key >= '1' && key <= '9' && best != null) {
            int predNum = Integer.parseInt(key + "");
            if (best.size() >= predNum) {
              sentence.append(best.get(predNum - 1).getWord() + " ");
              typeString = null;
              typeString = new StringBuilder();
              predictor.trainWord(best.get(predNum - 1).getWord());
              predictor.build();
            }
          }
        }
      }
      // Start redrawing onto screen.
      // Typed word set up and print.
      StdDraw.setPenColor(Color.BLACK);
      StdDraw.setFont(typeFont);
      StdDraw.text(width / 2, height / 4 * 3, typeString.toString());
      // Show prediction if it exists.
      if (best != null) {
        StdDraw.setPenColor(Color.BLUE);
        for (int i = 0; i < best.size(); i++) {
          StdDraw.text(
              width * ((i + .5) / best.size()),
              height / 2 + typeFont.getSize() / 2,
              best.get(i).getWord());
          if (multipleWords) {
            StdDraw.text(
                width * ((i + .5) / best.size()),
                height / 2 - typeFont.getSize() / 2,
                "" + (i + 1));
          }
        }
        StdDraw.setPenColor(Color.BLACK);
      }
      // Print out the sentence.
      StdDraw.setFont(sentenceFont);
      StdDraw.textRight(width - 20, height / 4, sentence.toString());
      // Pause the loop based on time set at beginning. Set to 0.
      StdDraw.show(pauseTime);
    }
  }