HeuristicTopDownClusteringTest.java

package test.kb.howtokb.clustering;

import java.util.List;

import kb.howtokb.clustering.HeuristicTopDownClustering;
import kb.howtokb.clustering.basicobj.CSKSimpleCluster;
import kb.howtokb.clustering.basicobj.Instance;
import kb.howtokb.taskframe.WikiHowTaskFrame;
import kb.howtokb.tools.InformationExtraction;

public class HeuristicTopDownClusteringTest {
	public static void main(String[] args) throws Exception {
		long startTime = System.currentTimeMillis();
//		Writer out = new BufferedWriter(new OutputStreamWriter(
//				new FileOutputStream("/var/tmp/cxchu/clustering-result/act-frame-domain-15k-heuristic-results.txt"), "utf-8"));
		
//		String input = "/var/tmp/cxchu/groundtruth-data/Old-Data/act-frame-sim.json";
//		String input = "/var/tmp/cxchu/clustering-result/bottomup-results/0-1k/bottom-up-cluster-56.json";
		
		String input = "howtokb-data/task-frame-before-clustering.json";
		List<WikiHowTaskFrame> allframe = InformationExtraction.getAllFrame(input);

		System.out.println("Initializing simple topdown clustering.....");
		
//		InstanceInLeafSimpleStopping<ActivityFrame> stopper = new InstanceInLeafSimpleStopping<>(5);
		HeuristicTopDownClustering cluster = new HeuristicTopDownClustering(allframe, false, 0.9, 5);
		
		System.out.println("Done! Start clustering.......");
		int k=2;
		List<CSKSimpleCluster<WikiHowTaskFrame>> res = cluster.splitACluster(cluster.getInputCluster(), k);
		
		System.out.println("Done! Results..............");
//		out.write("Number of clusters: " + res.size() + "\n");
		for (int i=0; i<res.size(); i++){
			List<Instance<WikiHowTaskFrame>> frames = res.get(i).getClusterMembers();
			for (int j=0; j<frames.size(); j++){
				System.out.println(frames.get(j).getData());
//				out.write(frames.get(j).getData() + "\n");
			}
			System.out.println(frames.size());
			System.out.println("====================================");
//			out.write("===========================\n");
		}
		
		System.out.println("Number of clusters: " + res.size());
		long endTime   = System.currentTimeMillis();
		long totalTime = (endTime - startTime)/1000;
		System.out.println("Running time: " + totalTime + "second(s)");
//		out.write("Running time: " + totalTime + " minute(s)");
//		out.close();
	}
}