associationmartix.java

自己写的search engine, 有 boolean search, fuzzy search

package searchingEngine.queryExpansion;


import java.util.*;

import searchingEngine.dataPreprocessing.invertedFile.DocNode;
import searchingEngine.dataPreprocessing.invertedFile.TermNode;
import searchingEngine.dataPreprocessing.wordPosition.DocNodeWpos;
import searchingEngine.utilites.dataConverter.RawConverter;
import searchingEngine.utilites.sorting.GetSortedList;


public class AssociationMartix {

	/**
	 * @param args
	 */
	public static final boolean IS_METRIC = true;
	public static final boolean IS_NOT_METRIC = false; 
	public final int termAmt;
	public final int termAmt2;
	private HeadNode[] assoMat;
	
	public AssociationMartix(int termAmt){
		this.termAmt = termAmt;
		termAmt2 = -1; 
		assoMat = new HeadNode[termAmt];
	}
	
	public AssociationMartix(int termAmt,int termAmt2){
		this.termAmt = termAmt;
		this.termAmt2 = termAmt2;
		assoMat = new HeadNode[termAmt];
	}
	
	public HeadNode[] getAssoMat(){
		return assoMat;
	}
	
	private int getFrequence(DocNodeWpos docNodeWpos) {
		return docNodeWpos.wpos_list.size();
	}

	private int getFrequence(TermNode termNode) {
		int result=0;
		for (int i =0;i<termNode.doc_list.size();i++)
			result += getFrequence((DocNodeWpos)termNode.doc_list.get(i));
		return result;
	}
	
	private int Cij(TermNode term0,TermNode term1) {
		int term0Index = 0,term1Index = 0;
		LinkedList<DocNode> term0doc_list = term0.doc_list;
		LinkedList<DocNode> term1doc_list = term1.doc_list;
		int term0DocLength = term0doc_list.size();
		int term1DocLength = term1doc_list.size();
		int correlation=0;
		DocNodeWpos info0;
		DocNodeWpos info1;
		while (term0Index<term0DocLength && term1Index<term1DocLength) {
			info0 = (DocNodeWpos)term0doc_list.get(term0Index);
			info1 = (DocNodeWpos)term1doc_list.get(term1Index);
			if (info0.compareTo(info1)==0) {
				correlation += getFrequence(info0) * getFrequence(info1);
				term0Index++;
				term1Index++;
			} else if (info0.compareTo(info1)<0) {
				term0Index++;
			} else term1Index++;			
		}
		return correlation;
	}
	
	private int r(int wpos0, int wpos1) {
		return Math.abs(wpos0-wpos1);
	}
	
	private float inverseR (int wpos0, int wpos1) {
		return (float)1.0/(float)r(wpos0,wpos1);
	}
	
	private float Mij(TermNode term0,TermNode term1) {
		int term0Index = 0,term1Index = 0;
		LinkedList<DocNode> term0DocNode = term0.doc_list;
		LinkedList<DocNode> term1DocNode = term1.doc_list;
		int term0DocLength = term0DocNode.size();
		int term1DocLength = term1DocNode.size();
		float correlation=0;
		float subCor = 0;
		DocNodeWpos info0;
		DocNodeWpos info1;
		while (term0Index<term0DocLength && term1Index<term1DocLength) {
			info0 = (DocNodeWpos)term0DocNode.get(term0Index);
			info1 = (DocNodeWpos)term1DocNode.get(term1Index);
			if (info0.compareTo(info1)==0) {
				for (int i=0; i<info0.wpos_list.size(); i++) {
					for (int j=0; j<info1.wpos_list.size(); j++) {
						subCor+= inverseR(info0.wpos_list.get(i),info1.wpos_list.get(j));					
					}
				}
				correlation += subCor;
				subCor =0;
				term0Index++;
				term1Index++;
			} else if (info0.compareTo(info1)<0) {
				term0Index++;
			} else term1Index++;			
		}
		return correlation;
	}
	private float Sij(float Cii, float Cjj, float Cij){
		return Cij/(Cii+Cjj-Cij);
	}

	private float Sij( TermNode Vi, TermNode Vj, float Mij){
		return Mij/(getFrequence(Vi)*getFrequence(Vj));
	}
	
	private void normalize() {
			for (int i=0;i<termAmt;i++)
				for (int j=0;j<i;j++) {
					setValueAt(Sij(getValueAt(i,i),getValueAt(j,j),getValueAt(i,j)),i,j);
				}
			for (int i=0;i<termAmt;i++)
				setValueAt(Sij(getValueAt(i,i),getValueAt(i,i),getValueAt(i,i)),i,i);
		}

	private void normalize(List<TermNode> termList) {
		for (int i=0;i<termAmt;i++) {
			for (int j=0;j<=i;j++) {
				setValueAt(Sij(termList.get(i),termList.get(j),getValueAt(i,j)),i,j);
			}
		}
	}
	
	public void generateAssoMat(List<TermNode> list, boolean isMetric){
		float correlations[];
		for (int i=0;i<termAmt;i++) {
			correlations = new float[i+1];
			for (int j=0;j<=i;j++) {
				if (isMetric) correlations[j] = Mij(list.get(i),list.get(j));
					else correlations[j] = Cij(list.get(i),list.get(j));
			}
			assoMat[i] = new HeadNode(list.get(i).term,correlations);
		}
		if (isMetric) normalize(list);
			else normalize();
	}
	
	public void generateAssoMat(List<TermNode> list,List<TermNode> list2, boolean isMetric){
		float correlations[];
		for (int i=0;i<termAmt;i++) {
			correlations = new float[termAmt2];
			for (int j=0;j<=termAmt2;j++) {
				if (isMetric) correlations[j] = Mij(list.get(i),list2.get(j));
					else correlations[j] = Cij(list.get(i),list2.get(j));
			}
			assoMat[i] = new HeadNode(list.get(i).term,correlations);
		}
		if (isMetric) normalize(list);
			else normalize();
	}
	
	public void setValueAt(float value, int i, int j) {
		assoMat[i].correlations[j] = value;
	}
	
	public float getValueAt(int i, int j) {
		return assoMat[i].correlations[j];
	}
	
	public String toString(){
		if (assoMat == null) return "empty";
		String result = "";
		for(int i =0 ; i<assoMat.length ; i++) {
			result += assoMat[i] + "\n"; 
		}
		return result;
	}
	
	public static void main(String[] args) throws Exception{
		//LinkedList<LinkedList<TermNode>> hash_table = (LinkedList<LinkedList<TermNode>>)RawConverter.loadObject("hashtable");
		LinkedList<TermNode> termList = (LinkedList<TermNode>)RawConverter.loadObject("G:/ir/wpos/X/combined0.dat");
		AssociationMartix am = new AssociationMartix(termList.size());
		am.generateAssoMat(termList,IS_METRIC);
		System.out.println("AssociationMartix:");
		RawConverter.saveObject(am,"G:/ir/wpos/X/am0.dat");
	}
	
	public class HeadNode implements Comparable<HeadNode>{
		public final String term;
		private float correlations[];
		
		public HeadNode(String term,float[] correlations){
			this.term = term;
			this.correlations = correlations;
		}

		public int compareTo(String arg0) {
			return term.compareToIgnoreCase(arg0);
		}
		public int compareTo(HeadNode arg0) {
			return term.compareToIgnoreCase(arg0.term);
		}
		
		public String toString(){
			String result = term + " " + " [ ";
			for (int i=0; i<correlations.length; i++){
				result += correlations[i] + " ";
			}
			result += "]";
			return result;
		}
	}
}