frequency.java

spam filter for email classifacation

package tc;
import java.io.*;
import java.lang.*;
import java.util.*;

public class Frequency {

	HashMap words;
	
	public Frequency(String fileIn)
	 {
	 	String text;
	 	try{
		FileReader fr = new FileReader(fileIn);
		BufferedReader br = new BufferedReader(fr);
		
		ArrayList lst=new ArrayList();
		String v[]=new String[1000];
		int n=1;
	//	try
	//		{
			while ((text=br.readLine())!= null)
				{
				StringTokenizer line= new StringTokenizer(text,".,* ()");
				try
					{
					while (line.hasMoreTokens())
						{
						// read word from line and stem word
							String word = new String();
							word = line.nextToken();
														
							//if(lst.contains(word))
							//	{
								//	System.out.print("\n"+word);
									v[n]=word;
									n++;									
							//	}		
								
							lst.add(word);
						}
				//	bw.newLine();
					}
				catch(Exception e)
					{
					System.err.println(e);
					}
				}
				
				  
			    Hashtable htb=new Hashtable();
				 
				 int s;				
			
				 
				for(int i=1;i<n;i++)
				  { 	
				  		s=0;
				  		for(int j=i+1;j<=n;j++)
				  			{
				  	 			if(v[i].equalsIgnoreCase(v[j]) && !v[i].equals(""))
				  	 				{
				  	 					v[j]="";
				  	 					s++;					  	 					
				  	 				}
				  	 		}
				  	 		
				  	 	htb.put(v[i],s+1);
				  } 
				  
				  
				  
				  Enumeration e=htb.keys();
				  
				  while(e.hasMoreElements())
				  	{
				  		String key = (String) e.nextElement();
      					System.out.println(key + " - " + htb.get(key));	
				  	}
			  
			}
			
			catch(Exception e){}
	  }
	  
    public static void main(String args[])
    	{
    		Frequency tcp=new Frequency("test.txt");    		    	
    	}
    	
} // end class