ActivityWordCategorySim.java

package clustering.sim;

import java.io.BufferedReader;
import java.io.FileReader;

import org.json.simple.JSONObject;
import org.json.simple.parser.JSONParser;

import activity.ActivityFrame;
import activity.ActivityWordsCategory;
import activity.extraction.JsonToActivityFrame;
import clustering.sim.w2v.Word2VecRunner;
import gnu.trove.TLongHashSet;

public class ActivityWordCategorySim {

	/*
	public static void main(String[] args) throws Exception {
		SparseSims distSparse = new SparseSims(0.5f);
		// need: ids of activity frames and ActivityWordsCategory
		//Test load all activity frame
		ActivityWordsCategory[] arr = new ActivityWordsCategory[1292250];
		
		String inputfile = "/var/tmp/cxchu/data-wordnet/act-frame.json";
		JSONParser parser = new JSONParser();

		try (BufferedReader br = new BufferedReader(new FileReader(inputfile))) {

			String sCurrentLine;
			int i=0;
			
			while ((sCurrentLine = br.readLine()) != null) {
				
				Object obj = parser.parse(sCurrentLine);
				JSONObject jsonObject = (JSONObject) obj;
				ActivityFrame newframe = JsonToActivityFrame.jsonToActivityFrame(jsonObject);
				ActivityWordsCategory tmp = new ActivityWordsCategory(newframe.getID(), 
						Integer.parseInt(newframe.getActivity().getCategoryID()), newframe.getActivity().getVerb() + ";" + newframe.getActivity().getObject());
				System.out.println(i);
				arr[i++] = tmp;
			}
		}
		
		for (int i = 0; i < arr.length; i++){
			ActivityWordsCategory e1 = arr[i];
			// Only store the upper triangular matrix.
			for (int j = i + 1; j < arr.length; j++)
					distSparse.set(i, j, (float) Word2VecRunner.simPair(e1.getActivityStrong(), arr[j].getActivityStrong()));
		}
		System.out.println("Done!");
		
		
	}
	*/

	/**
	 * Only stores the upper triangular matrix.
	 * 
	 * @author cxchu
	 *
	 */
	public static class SparseSims {

		private TLongHashSet simPairs;
		private float threshold;

		public SparseSims(float threshold) {
			simPairs = new TLongHashSet();
			this.threshold = threshold;
		}

		public void set(int x, int y, float value) {
			if (value < threshold)
				return;
			long key = intpairToLong(x, y);
			simPairs.add(key);
		}

		public boolean get(int x, int y) {
			long key = intpairToLong(x, y);
			if (simPairs.contains(key))
				return true;

			return simPairs.contains(intpairToLong(y, x));
		}

		private long intpairToLong(int l, int r) {
			return ((long) l << 32) + r;
		}

	}

}