图模型与概率统计assignment02

问题描述

原谅我不放文字了。（我估计没人看这个博客吧，当自己笔记用了，有价值的话后期再改）

代码

注释就加了几行

import java.io.BufferedReader;
import java.io.File;
import java.io.FileReader;
import java.util.Random;

public class SeussAndSaki {

	final static String alphabet = "abcdefghijklmnopqrstuvwxyz";

	public static void main(String[] args) throws Exception {
		String fileA = "spamiam.txt";
		String fileB = "saki_story.txt";
		randomA();
		System.out.println("\nspamiam.txt in strategy B");randomB(fileA);
		System.out.println("\nspamiam.txt in strategy C");randomC(fileA);
		System.out.println("\nspamiam.txt in strategy D");randomD(fileA);
		System.out.println("\nsaki_story.txt in strategy B");randomB(fileB);
		System.out.println("\nsaki_story.txt in strategy C");randomC(fileB);
		System.out.println("\nsaki_story.txt in strategy D");randomD(fileB);
	}

	/* Completely Random */
	public static void randomA() {
		Random random = new Random();
		for (int i = 1; i <= 10; i++) {
			for (int j = 1; j <= 10; j++) {
				for (int k = 1; k <= 4; k++) {
					System.out.print(alphabet.charAt(random.nextInt(26)));
				}
				System.out.print(" ");
			}
			System.out.println();
		}
	}
	
	/* Random with frequency as probability */
	public static void randomB(String fileName) throws Exception {
		int[] frequency = new int[26];
		String[] splitedTxt = readToString(fileName);
		/* Initial frequency */
		for(int i=0; i<26; i++)
			frequency[i] = 0;
		/* Calculate frequency */
		for(String s : splitedTxt)
			for(int i=0; i<s.length(); i++)
				frequency[s.charAt(i)-97]++;
		/* Sum frequency for Random with frequency */
		for(int i=1; i<=25; i++)
			frequency[i] += frequency[i-1];
		for (int i = 1; i <= 10; i++) {
			for (int j = 1; j <= 10; j++) {
				for (int k = 1; k <= 4; k++) {
					System.out.print(strategyB(frequency));
				}
				System.out.print(" ");
			}
			System.out.println();
		}
	}
	
	/* Random with one-step transition probability */
	public static void randomC(String fileName) throws Exception {
		int[][] frequency = new int[26][27];
		int[] total = new int[26];
		String[] splitedTxt = readToString(fileName);
		/* Initial one-step TP matrix */
		for(int i=0; i<26; i++)
			for(int j=0; j<27; j++)
				frequency[i][j] = 0;
		initTotal(total);
		/* Calculate matrix and total */
		for(String s : splitedTxt) {
			if(s.length() == 0)
				continue;
			total[s.charAt(0)-97]++;
			for(int i=1; i<s.length(); i++) {
				char cn1 = s.charAt(i-1);
				char cn2 = s.charAt(i);
				frequency[cn1-97][cn2-97]++;
				total[cn2-97]++;
			}
		}
		/* Sum total and matrix for Random with transition frequency */
		sumTotal(total);
		for(int i=0; i<26; i++)
			for(int j=1; j<=26; j++)
				frequency[i][j] += frequency[i][j-1];
		/* Generate words */
		for (int i = 1; i <= 10; i++) {
			for (int j = 1; j <= 10; j++) {
				char pre = strategyB(total);
				System.out.print(pre);
				for(int k=1; k<=3; k++) {
					char ch = strategyC(total, frequency, pre);
					System.out.print(ch);
					pre = ch;
				}
				System.out.print(" ");
			}
			System.out.println();
		}
	}
	
	/* Random with two-step transition probability  */
	public static void randomD(String fileName) throws Exception {
		Random random = new Random();
		int frequency[][][] = new int[26][26][27];
		int frequencyC[][] = new int[26][27];
		int[] total = new int[26];
		String[] splitedTxt = readToString(fileName);
		/* Initial two-step TP matrix */
		for(int i=0; i<26; i++)
			for(int j=0; j<26; j++)
				for(int k=0; k<27; k++)
					frequency[i][j][k] = 0;
		/* Initial one-step TP matrix */
		for(int i=0; i<26; i++)
			for(int j=0; j<27; j++)
				frequencyC[i][j] = 0;
		initTotal(total);
		/* Calculate the one-step TP matrix and two-step TP matrix and total */
		for(String s : splitedTxt) {
			if(s.length() == 0)
				continue;
			else if(s.length() == 1) {
				total[s.charAt(0)-97]++;
				continue;
			}
			total[s.charAt(0)-97]++;
			total[s.charAt(1)-97]++;
			for(int i=2; i<s.length(); i++) {
				char cn1 = s.charAt(i-2);
				char cn2 = s.charAt(i-1);
				char cn3 = s.charAt(i);
				frequency[cn1-97][cn2-97][cn3-97]++;
				frequencyC[cn1-97][cn2-97]++;
				total[cn3-97]++;
			}
			frequencyC[s.charAt(s.length()-2)-97][s.charAt(s.length()-1)-97]++;
		}
		/* Sum the matrices and total for Random with transition frequency */
		sumTotal(total);
		for(int i=0; i<26; i++) {
			for(int j=0; j<26; j++) {
				for(int k=1; k<27; k++) {
					frequency[i][j][k] += frequency[i][j][k-1];
				}
			}
		}
		for(int i=0; i<26; i++) {
			for(int j=1; j<27; j++) {
				frequencyC[i][j] += frequencyC[i][j-1];
			}
		}
		/* Generate words */
		for (int i = 1; i <= 10; i++) {
			for (int j = 1; j <= 10; j++) {
				char cn1 = strategyB(total);
				char cn2 = strategyC(total, frequencyC, cn1);
				System.out.print(cn1);
				System.out.print(cn2);
				for(int k=1; k<=2; k++) {
					if(frequency[cn1-97][cn2-97][26] <= 0) {
						cn1 = cn2;
						cn2 = strategyC(total, frequencyC, cn1);
						System.out.print(cn2);
					}else {
						int r = random.nextInt(frequency[cn1-97][cn2-97][26]);
						for(int m=0; m<26; m++) {
							if(r <= frequency[cn1-97][cn2-97][m]) {
								cn1 = cn2;
								cn2 = alphabet.charAt(m);
								System.out.print(cn2);
								break;
							}
						}
					}
				}
				System.out.print(" ");
			}
			System.out.println();
		}
	}
	
	/* Read files and ignore all characters but a-z(A-Z), and return String[] */
	public static String[] readToString(String filename) throws Exception {
		File file = new File(filename);
		FileReader reader = new FileReader(file);
		BufferedReader bf = new BufferedReader(reader);
		String txt = "";
		String line = null;
		while((line = bf.readLine()) != null) {
			txt += (line + " ");
		}
		txt = txt.toLowerCase();
		String[] splitedTxt = txt.split(" ");
		for(int i=0; i<splitedTxt.length; i++) {
			splitedTxt[i] = splitedTxt[i].replaceAll("[^a-zA-Z]", "");
		}
		bf.close();
		return splitedTxt;
	}
	/* Generate a character with frequency as probability */
	public static char strategyB(int[] frequency) {
		Random random = new Random();
		int r = random.nextInt(frequency[25]);
		for(int m=0; m<26; m++) {
			if(r <= frequency[m]) {
				return alphabet.charAt(m);
			}
		}
		return alphabet.charAt(25);
	}
	/* Generate a character with one-step transition probability of char pre */
	public static char strategyC(int[] total, int[][] frequency, char pre) {
		Random random = new Random();
		if(frequency[pre-97][26] <= 0) {
			pre = strategyB(total);
			return pre;
		}else {
			int r = random.nextInt(frequency[pre-97][26]);
			for(int m=0; m<26; m++) {
				if(r <= frequency[pre-97][m]) {
					return alphabet.charAt(m);
				}
			}
		}
		return alphabet.charAt(25);
	}
	/* Initialize int[] total */
	public static void initTotal(int[] total) {
		for(int i=0; i<26; i++)
			total[i] = 0;
	}
	/* Sum int[] total */
	public static void sumTotal(int[] total) {
		for(int i=1; i<=25; i++)
			total[i] += total[i-1];
	}
}