technicalinformation.java

Java 编写的多种数据挖掘算法 包括聚类、分类、预处理等

    return generateID();
  }
  
  /**
   * sets the value for the given field, overwrites any previously existing one.
   * 
   * @param field	the field to set the value for
   * @param value	the value of the field
   */
  public void setValue(Field field, String value) {
    m_Values.put(field, value);
  }
  
  /**
   * returns the value associated with the given field, or empty if field is
   * not currently stored.
   * 
   * @param field	the field to retrieve the value for
   * @return		the value associated with this field, empty if not existing
   */
  public String getValue(Field field) {
    if (m_Values.containsKey(field))
      return (String) m_Values.get(field);
    else
      return "";
  }
  
  /**
   * returns TRUE if the field is stored and has a value different from the
   * empty string.
   * 
   * @param field	the field to check
   * @return 		true if a value is stored for the field and non-empty
   */
  public boolean exists(Field field) {
    return (    m_Values.containsKey(field) 
	     && (((String) m_Values.get(field)).length() != 0) );
  }
  
  /**
   * returns an enumeration over all the stored fields
   * 
   * @return		all currently stored fields
   */
  public Enumeration fields() {
    return m_Values.keys();
  }
  
  /**
   * returns true if there are further technical informations stored in this
   * 
   * @return true if there are further technical informations available
   */
  public boolean hasAdditional() {
    return (m_Additional.size() > 0);
  }
  
  /**
   * returns an enumeration of all the additional technical informations (if
   * there are any)
   * 
   * @return an enumeration over all additional technical informations
   */
  public Enumeration additional() {
    return m_Additional.elements();
  }
  
  /**
   * adds the given information to the list of additional technical 
   * informations
   * 
   * @param value the information to add
   */
  public void add(TechnicalInformation value) {
    if (value == this)
      throw new IllegalArgumentException("Can't add object to itself!");
    m_Additional.add(value);
  }
  /**
   * Adds an empty technical information with the given type to the list
   * of additional informations and returns the instance.
   * 
   * @param type	the type of the new information to add
   * @return 		the generated information
   */
  public TechnicalInformation add(Type type) {
    TechnicalInformation 	result;
    
    result = new TechnicalInformation(type);
    add(result);
    
    return result;
  }
  
  /**
   * Returns a plain-text string representing this technical information.
   * Note: it only returns a string based on some fields. At least AUTHOR,
   * YEAR and TITLE are necessary.
   * 
   * @return		a string representation of this information
   */
  public String toString() {
    String	result;
    String[]	authors;
    int		i;
    Enumeration	enm;
    
    result  = "";
    authors = getAuthors();
    
    // BOOK
    if (getType() == Type.BOOK) {
      for (i = 0; i < authors.length; i++) {
	if (i > 0)
	  result += ", ";
	result += authors[i];
      }
      result += " (" + getValue(Field.YEAR) + ").";
      result += " " + getValue(Field.TITLE) + ".";
      result += " " + getValue(Field.PUBLISHER);
      if (exists(Field.ADDRESS))
	result += ", " + getValue(Field.ADDRESS);
      result += ".";
    }
    // ARTICLE
    else if (getType() == Type.ARTICLE) {
      for (i = 0; i < authors.length; i++) {
	if (i > 0)
	  result += ", ";
	result += authors[i];
      }
      result += " (" + getValue(Field.YEAR) + ").";
      result += " " + getValue(Field.TITLE) + ".";
      
      // journal
      if (exists(Field.JOURNAL)) {
	result += " " + getValue(Field.JOURNAL) + ".";

	if (exists(Field.VOLUME))
	  result += " " + getValue(Field.VOLUME);
	if (exists(Field.NUMBER))
	  result += "(" + getValue(Field.NUMBER) + ")";
	if (exists(Field.PAGES))
	  result += ":" + getValue(Field.PAGES);
	
	result += ".";
      }
      
      // other than JOURNAL???

      // URL
      if (exists(Field.URL))
	result += " URL " + getValue(Field.URL) + ".";
    }
    // CONFERENCE/INPROCEEDINGS
    else if ( (getType() == Type.CONFERENCE) || (getType() == Type.INPROCEEDINGS) ) {
      for (i = 0; i < authors.length; i++) {
	if (i > 0)
	  result += ", ";
	result += authors[i];
      }
      result += ": " + getValue(Field.TITLE) + ".";
      result += " In: " + getValue(Field.BOOKTITLE);
      
      if (exists(Field.ADDRESS))
	result += ", " + getValue(Field.ADDRESS);
      if (exists(Field.PAGES))
	result += ", " + getValue(Field.PAGES);
	
      result += ", " + getValue(Field.YEAR) + ".";
    }
    // INCOLLECTION
    else if (getType() == Type.INCOLLECTION) {
      for (i = 0; i < authors.length; i++) {
	if (i > 0)
	  result += ", ";
	result += authors[i];
      }
      result += ": " + getValue(Field.TITLE) + ".";
      result += " In ";
      if (exists(Field.EDITOR))
	result += getValue(Field.EDITOR) + ", editors, ";
      result += getValue(Field.BOOKTITLE);
      
      if (exists(Field.ADDRESS))
	result += ", " + getValue(Field.ADDRESS);
      if (exists(Field.PAGES))
	result += ", " + getValue(Field.PAGES);
	
      result += ", " + getValue(Field.YEAR) + ".";
    }
    // default
    else {
      for (i = 0; i < authors.length; i++) {
	if (i > 0)
	  result += ", ";
	result += authors[i];
      }
      result += " (" + getValue(Field.YEAR) + ").";
      result += " " + getValue(Field.TITLE) + ".";
      if (exists(Field.ADDRESS))
	result += " " + getValue(Field.ADDRESS) + ".";
      if (exists(Field.URL))
	result += " URL " + getValue(Field.URL) + ".";
    }
    
    // additional informations?
    enm = additional();
    while (enm.hasMoreElements()) {
      result += "\n\n" + enm.nextElement().toString();
    }
    
    return result;
  }
  
  /**
   * Returns a BibTex string representing this technical information. 
   * Note: this is just a very raw implementation, special characters need to
   * be escaped manually for LaTeX.
   * 
   * @return		the BibTeX representation of this information
   */
  public String toBibTex() {
    String		result;
    Enumeration		enm;
    Field		field;
    Vector		list;
    int			i;
    
    result = "@" + getType() + "{" + getID() + "";
    
    // sort the fields
    list = new Vector();
    enm  = fields();
    while (enm.hasMoreElements())
      list.add(enm.nextElement());
    Collections.sort(list);
    
    // list field=value pairs
    for (i = 0; i < list.size(); i++) {
      field = (Field) list.get(i);
      if (!exists(field))
	continue;
      result += ",\n   " + field + " = {" + getValue(field) +  "}";
    }
    
    result += "\n}";
    
    // additional informations?
    enm = additional();
    while (enm.hasMoreElements()) {
      result += "\n\n" + ((TechnicalInformation) enm.nextElement()).toBibTex();
    }
    
    return result;
  }
  
  /**
   * Prints some examples of technical informations if there are no 
   * commandline options given. Otherwise the information of a given 
   * TechnicalInformationHandler can be printed. <p/>
   * 
   * Valid options are: <p/>
   * 
   * -W classname <br/>
   *  The classname of the TechnicalInformationHandler to print the 
   *  information for <p/>
   *  
   * -bibtex <br/>
   *  Print the information in BibTeX format <p/>
   *  
   * -plaintext <br/>
   *  Print the information in plain text format <p/>
   * 
   * @param args 	the commandline options
   * @throws Exception	if the option parsing fails
   */
  public static void main(String[] args) throws Exception {
    TechnicalInformation	info;
    TechnicalInformation	additional;
    String			tmpStr;
    Class			cls;
    TechnicalInformationHandler	handler;
    
    // example from command line
    if (args.length != 0) {
      info = null;
      
      tmpStr = Utils.getOption('W', args);
      if (tmpStr.length() != 0) {
	cls     = Class.forName(tmpStr);
	handler = (TechnicalInformationHandler) cls.newInstance();
	info    = handler.getTechnicalInformation();
      }
      else {
	throw new IllegalArgumentException("A classname has to be provided with the -W option!");
      }
      
      if (Utils.getFlag("bibtex", args))
        System.out.println("\n" + handler.getClass().getName() + ":\n" + info.toBibTex());

      if (Utils.getFlag("plaintext", args))
        System.out.println("\n" + handler.getClass().getName() + ":\n" + info.toString());
    }
    else {
      // first example
      info = new TechnicalInformation(Type.BOOK);
      info.setValue(Field.AUTHOR, "Ross Quinlan");
      info.setValue(Field.YEAR, "1993");
      info.setValue(Field.TITLE, "C4.5: Programs for Machine Learning");
      info.setValue(Field.PUBLISHER, "Morgan Kaufmann Publishers");
      info.setValue(Field.ADDRESS, "San Mateo, CA");
      additional = info;
      
      System.out.println("\ntoString():\n" + info.toString());
      System.out.println("\ntoBibTex():\n" + info.toBibTex());
      
      // second example
      info = new TechnicalInformation(Type.INPROCEEDINGS);
      info.setValue(Field.AUTHOR, "Freund, Y. and Mason, L.");
      info.setValue(Field.YEAR, "1999");
      info.setValue(Field.TITLE, "The alternating decision tree learning algorithm");
      info.setValue(Field.BOOKTITLE, "Proceeding of the Sixteenth International Conference on Machine Learning");
      info.setValue(Field.ADDRESS, "Bled, Slovenia");
      info.setValue(Field.PAGES, "124-133");
      
      System.out.println("\ntoString():\n" + info.toString());
      System.out.println("\ntoBibTex():\n" + info.toBibTex());
      
      // third example
      info = new TechnicalInformation(Type.ARTICLE);
      info.setValue(Field.AUTHOR, "R. Quinlan");
      info.setValue(Field.YEAR, "1986");
      info.setValue(Field.TITLE, "Induction of decision trees");
      info.setValue(Field.JOURNAL, "Machine Learning");
      info.setValue(Field.VOLUME, "1");
      info.setValue(Field.NUMBER, "1");
      info.setValue(Field.PAGES, "81-106");
      
      additional = new TechnicalInformation(Type.BOOK);
      additional.setValue(Field.AUTHOR, "Ross Quinlan");
      additional.setValue(Field.YEAR, "1993");
      additional.setValue(Field.TITLE, "C4.5: Programs for Machine Learning");
      additional.setValue(Field.PUBLISHER, "Morgan Kaufmann Publishers");
      additional.setValue(Field.ADDRESS, "San Mateo, CA");
      info.add(additional);
      
      System.out.println("\ntoString():\n" + info.toString());
      System.out.println("\ntoBibTex():\n" + info.toBibTex());
    }
  }
}