complementnaivebayes.java

代码是一个分类器的实现,其中使用了部分weka的源代码。可以将项目导入eclipse运行

        
        return  "Class for building and using a Complement class Naive Bayes "+
                "classifier.\n\nFor more information see, \n\n"+
                getTechnicalInformation().toString() + "\n\n" +
                "P.S.: TF, IDF and length normalization transforms, as "+
                "described in the paper, can be performed through "+
                "weka.filters.unsupervised.StringToWordVector.";
    }

    /**
     * Returns an instance of a TechnicalInformation object, containing 
     * detailed information about the technical background of this class,
     * e.g., paper reference or book this class is based on.
     * 
     * @return the technical information about this class
     */
    public TechnicalInformation getTechnicalInformation() {
      TechnicalInformation 	result;
      
      result = new TechnicalInformation(Type.INPROCEEDINGS);
      result.setValue(Field.AUTHOR, "Jason D. Rennie and Lawrence Shih and Jaime Teevan and David R. Karger");
      result.setValue(Field.TITLE, "Tackling the Poor Assumptions of Naive Bayes Text Classifiers");
      result.setValue(Field.BOOKTITLE, "ICML");
      result.setValue(Field.YEAR, "2003");
      result.setValue(Field.PAGES, "616-623");
      result.setValue(Field.PUBLISHER, "AAAI Press");
      
      return result;
    }

    /**
     * Returns default capabilities of the classifier.
     *
     * @return      the capabilities of this classifier
     */
    public Capabilities getCapabilities() {
      Capabilities result = super.getCapabilities();

      // attributes
      result.enable(Capability.NUMERIC_ATTRIBUTES);
      result.enable(Capability.MISSING_VALUES);

      // class
      result.enable(Capability.NOMINAL_CLASS);
      result.enable(Capability.MISSING_CLASS_VALUES);
      
      return result;
    }
    
    /**
     * Generates the classifier.
     *
     * @param instances set of instances serving as training data 
     * @throws Exception if the classifier has not been built successfully
     */
    public void buildClassifier(Instances instances) throws Exception {

      // can classifier handle the data?
      getCapabilities().testWithFail(instances);

      // remove instances with missing class
      instances = new Instances(instances);
      instances.deleteWithMissingClass();
      
        numClasses = instances.numClasses();
	int numAttributes = instances.numAttributes();
        
        header = new Instances(instances, 0);
	double [][] ocrnceOfWordInClass = new double[numClasses][numAttributes];        
        wordWeights = new double[numClasses][numAttributes];
        //double [] docsPerClass = new double[numClasses];
	double[] wordsPerClass = new double[numClasses];
        double totalWordOccurrences = 0;
        double sumOfSmoothingParams = (numAttributes-1)*smoothingParameter;
        int classIndex = instances.instance(0).classIndex();        
	Instance instance;
	int docClass;
	double numOccurrences;        
        
        java.util.Enumeration enumInsts = instances.enumerateInstances();
	while (enumInsts.hasMoreElements()) {
		instance = (Instance) enumInsts.nextElement();
		docClass = (int)instance.value(classIndex);
		//docsPerClass[docClass] += instance.weight();
		
		for(int a = 0; a<instance.numValues(); a++)
		    if(instance.index(a) != instance.classIndex()) {
			    if(!instance.isMissing(a)) {
				    numOccurrences = instance.valueSparse(a) * instance.weight();
				    if(numOccurrences < 0)
					throw new Exception("Numeric attribute"+
                                                  " values must all be greater"+
                                                  " or equal to zero.");
                                    totalWordOccurrences += numOccurrences;
				    wordsPerClass[docClass] += numOccurrences;
				    ocrnceOfWordInClass[docClass]
                                          [instance.index(a)] += numOccurrences;
                                    //For the time being wordweights[0][i] 
                                    //will hold the total occurrence of word
                                    // i over all classes
                                    wordWeights[0]
                                          [instance.index(a)] += numOccurrences;
                            }
                    }
        }

	//Calculating the complement class probability for all classes except 0        
	for(int c=1; c<numClasses; c++) {
            //total occurrence of words in classes other than c
            double totalWordOcrnces = totalWordOccurrences - wordsPerClass[c];

            for(int w=0; w<numAttributes; w++) {
                if(w != classIndex ) {
                     //occurrence of w in classes other that c
                    double ocrncesOfWord = 
                                wordWeights[0][w] - ocrnceOfWordInClass[c][w];

                    wordWeights[c][w] = 
                        Math.log((ocrncesOfWord+smoothingParameter) / 
                                (totalWordOcrnces+sumOfSmoothingParams));
                }
            }
        }
        
	//Now calculating the complement class probability for class 0
        for(int w=0; w<numAttributes; w++) {
            if(w != classIndex) {
                //occurrence of w in classes other that c
                double ocrncesOfWord = wordWeights[0][w] - ocrnceOfWordInClass[0][w];
                //total occurrence of words in classes other than c
                double totalWordOcrnces = totalWordOccurrences - wordsPerClass[0];
                
                wordWeights[0][w] =
                Math.log((ocrncesOfWord+smoothingParameter) /
                (totalWordOcrnces+sumOfSmoothingParams));
            }            
        }
        
       	//Normalizing weights
        if(m_normalizeWordWeights==true)
            for(int c=0; c<numClasses; c++) {
                double sum=0;
                for(int w=0; w<numAttributes; w++) {
                    if(w!=classIndex)
                        sum += Math.abs(wordWeights[c][w]);
                }
                for(int w=0; w<numAttributes; w++) {
                    if(w!=classIndex) {
                        wordWeights[c][w] = wordWeights[c][w]/sum;
                    }
                }
            }

    }

    
    /**
     * Classifies a given instance. <p>
     *
     * The classification rule is: <br>
     *     MinC(forAllWords(ti*Wci)) <br>
     *      where <br>
     *         ti is the frequency of word i in the given instance <br>
     *         Wci is the weight of word i in Class c. <p>
     *
     * For more information see section 4.4 of the paper mentioned above
     * in the classifiers description.
     *
     * @param instance the instance to classify
     * @return the index of the class the instance is most likely to belong.
     * @throws Exception if the classifier has not been built yet.
     */
    public double classifyInstance(Instance instance) throws Exception {

        if(wordWeights==null)
            throw new Exception("Error. The classifier has not been built "+
                                "properly.");
        
        double [] valueForClass = new double[numClasses];
	double sumOfClassValues=0;
	
	for(int c=0; c<numClasses; c++) {
	    double sumOfWordValues=0;
	    for(int w=0; w<instance.numValues(); w++) {
                if(instance.index(w)!=instance.classIndex()) {
                    double freqOfWordInDoc = instance.valueSparse(w);
                    sumOfWordValues += freqOfWordInDoc * 
                                  wordWeights[c][instance.index(w)];
                }
	    }
	    //valueForClass[c] = Math.log(probOfClass[c]) - sumOfWordValues;
	    valueForClass[c] = sumOfWordValues;
	    sumOfClassValues += valueForClass[c];
	}

        int minidx=0;
	for(int i=0; i<numClasses; i++)
	    if(valueForClass[i]<valueForClass[minidx])
		minidx = i;
	
	return minidx;
    }


    /**
     * Prints out the internal model built by the classifier. In this case
     * it prints out the word weights calculated when building the classifier.
     */
    public String toString() {
        if(wordWeights==null) {            
            return "The classifier hasn't been built yet.";
        }
        
        int numAttributes = header.numAttributes();
        StringBuffer result = new StringBuffer("The word weights for each class are: \n"+
                                               "------------------------------------\n\t");
        
        for(int c = 0; c<numClasses; c++)
            result.append(header.classAttribute().value(c)).append("\t");
        
        result.append("\n");
        
        for(int w = 0; w<numAttributes; w++) {
            result.append(header.attribute(w).name()).append("\t");
            for(int c = 0; c<numClasses; c++)
                result.append(Double.toString(wordWeights[c][w])).append("\t");
            result.append("\n");
        }
        
        return result.toString();
    }
    
    
    /**
     * Main method for testing this class.
     *
     * @param argv the options
     */
    public static void main(String [] argv) {
      runClassifier(new ComplementNaiveBayes(), argv);
    }        
}