SimplePruningSimilarity.java

package kb.howtokb.clustering;

import java.io.FileNotFoundException;
import java.io.IOException;
import java.sql.SQLException;
import java.util.Map;
import java.util.Set;

import edu.stanford.nlp.util.Pair;
import kb.howtokb.clustering.basicobj.ActivityWordsCategory;
import kb.howtokb.clustering.sim.CategorySimilarity;
import kb.howtokb.clustering.sim.Coefficient;
import kb.howtokb.clustering.sim.w2v.Word2VecSimilarity;
import kb.howtokb.utils.BijectiveMap;
import kb.howtokb.utils.FileLines;
import kb.howtokb.utils.IDMap;

public class SimplePruningSimilarity implements ISimilarity<ActivityWordsCategory>{

	private static CategorySimilarity cate;
	private double threshold;
	private String model;
	private String allAct;
	private Set<String> allActList;
	private ActivityCachedSim aSim;
	public SimplePruningSimilarity(double threshold, String model, String allAct) throws SQLException, IOException, ClassNotFoundException {
		cate = new CategorySimilarity();
		
		this.threshold = threshold;
		this.model = model;
		this.allAct = allAct;
		ISimilarity<String> word2vecSim = new Word2VecSimilarity(this.model, 0, false);
		IDMap<String, Integer> vIDs = new IDMap<>(0);
		IDMap<String, Integer> nIDs = new IDMap<>(0);
		BijectiveMap<String, Pair<Integer, Integer>> aIDs = new BijectiveMap<>();
		loadIDs(this.allAct, vIDs, nIDs, aIDs);

		//Using separate threshold for verb and noun v: 0.747, n: 0.67
		//Or can use same threshold, the bottom neck: vvnn: 0.5
		ActivityComponentSim vSim = new ActivityComponentSim(Coefficient.V_THRES, vIDs, word2vecSim);
		ActivityComponentSim nSim = new ActivityComponentSim(Coefficient.O_THRES, nIDs, word2vecSim);
		this.aSim = new ActivityCachedSim(aIDs, vSim, nSim);
	}
	
	public SimplePruningSimilarity(double threshold, String model, Set<String> allAct) throws SQLException, IOException, ClassNotFoundException {
		cate = new CategorySimilarity();
		
		this.threshold = threshold;
		this.model = model;
		this.allActList = allAct;
		ISimilarity<String> word2vecSim = new Word2VecSimilarity(this.model, 0, false);
		IDMap<String, Integer> vIDs = new IDMap<>(0);
		IDMap<String, Integer> nIDs = new IDMap<>(0);
		BijectiveMap<String, Pair<Integer, Integer>> aIDs = new BijectiveMap<>();
		loadIDs(this.allActList, vIDs, nIDs, aIDs);

		//Using separate threshold for verb and noun v: 0.747, n: 0.67
		//Or can use same threshold, the bottom neck: vvnn: 0.5
		ActivityComponentSim vSim = new ActivityComponentSim(Coefficient.V_THRES, vIDs, word2vecSim);
		ActivityComponentSim nSim = new ActivityComponentSim(Coefficient.O_THRES, nIDs, word2vecSim);
		this.aSim = new ActivityCachedSim(aIDs, vSim, nSim);
	}
	
	@Override
	public double sim(ActivityWordsCategory e1, ActivityWordsCategory e2) {
		// TODO we don't have function to compute in this case
		return 0;
	}

	@Override
	public Map<ActivityWordsCategory, Double> getNeighbors(ActivityWordsCategory e) {
		// TODO We don't need this function right now
		return null;
	}
	/**
	 * Check similar between two activities in a simple way
	 * if similarity between two category is less than a threshold
	 * and similarity between two strong activities is less than a threshold
	 * then they are dissimilar, otw not sure and return true
	 * @param two ActivityWordsCategory objects
	 * @return true/false
	 */
	@Override
	public boolean simThreshold(ActivityWordsCategory e1, ActivityWordsCategory e2, double minthreshold) throws Exception {
		if (!cate.isSim(e1.getCatID(), e2.getCatID()))
			if (!aSim.sim(e1.getActivityStrong(), e2.getActivityStrong()))
				return false;
		return true;
			
	}
	/**
	 * loadID from a file including a list of activities
	 * @param activityList
	 * @param vIDs
	 * @param nIDs
	 * @param aIDs
	 * @throws FileNotFoundException
	 */
	private static void loadIDs(String activityList, IDMap<String, Integer> vIDs, IDMap<String, Integer> nIDs,
			BijectiveMap<String, Pair<Integer, Integer>> aIDs) throws FileNotFoundException {
		for (String a : new FileLines(activityList)) {
			// return,from,work
			String[] vn = a.split(";");
			String verb = "v_" + (vn[0].contains(" ")?vn[0].split(" ")[0]:vn[0]);
			String noun = "n_" + (vn[1].contains(" ")?vn[1].split(" ")[vn[1].split(" ").length - 1]:vn[1]);
			int vid = vIDs.getAvailableGlobalID();
			vIDs.add(verb);
			int nid = nIDs.getAvailableGlobalID();
			nIDs.add(noun);
			aIDs.put(a, new Pair<Integer, Integer>(vid, nid));
		}
		System.out.println("Load map verb to id done! Size: " + vIDs.size());
		System.out.println("Load map noun to id done! Size: " + nIDs.size());
	}
	
	/**
	 * loadID from list of activities
	 * @param activityList
	 * @param vIDs
	 * @param nIDs
	 * @param aIDs
	 * @throws FileNotFoundException
	 */
	private static void loadIDs(Set<String> activityList, IDMap<String, Integer> vIDs, IDMap<String, Integer> nIDs,
			BijectiveMap<String, Pair<Integer, Integer>> aIDs) throws FileNotFoundException {
		for (String a : activityList) {
			// return,from,work
			String[] vn = a.split(";");
			String verb = "v_" + (vn[0].contains(" ")?vn[0].split(" ")[0]:vn[0]);
			String noun = "n_" + (vn[1].contains(" ")?vn[1].split(" ")[vn[1].split(" ").length - 1]:vn[1]);
			int vid = vIDs.getAvailableGlobalID();
			vIDs.add(verb);
			int nid = nIDs.getAvailableGlobalID();
			nIDs.add(noun);
			aIDs.put(a, new Pair<Integer, Integer>(vid, nid));
		}
		System.out.println("Load map verb to id done! Size: " + vIDs.size());
		System.out.println("Load map noun to id done! Size: " + nIDs.size());
	}

}