HeuristicBottomUpClusteringTest.java

package kb.howtokb.clustering;

import java.io.BufferedWriter;
import java.io.FileOutputStream;
import java.io.OutputStreamWriter;
import java.io.Writer;
import java.util.List;

import kb.howtokb.clustering.HeuristicBottomupClustering.ActivitySuperCluster;
import kb.howtokb.clustering.sim.Coefficient;
import kb.howtokb.taskframe.WikiHowTaskFrame;
import kb.howtokb.tools.InformationExtraction;

public class HeuristicBottomUpClusteringTest {
	public static void main(String[] args) throws Exception {
		
		long startTime = System.currentTimeMillis();
		
		 
		String activityTb = "all-words-category.txt";
		HeuristicBottomupClustering cluster = new HeuristicBottomupClustering(activityTb);
		
		double threshold = Coefficient.VVNN_TRHES;
		String model = "articles-word2vec-word-pos.model.txt";
		String allAct = "resources/all-strong-activities.txt";
		SimplePruningSimilarity simFunc = new SimplePruningSimilarity(threshold, model, allAct);
		List<ActivitySuperCluster> results = cluster.cluster(simFunc, Coefficient.VVNN_TRHES);
		System.out.println("Number of clusters: " + results.size());
		String output = "/var/tmp/cxchu/clustering-result-wikihow-task/bottom-up-cluster-";
		
		String input = "/var/tmp/cxchu/data-server/For-Database/act-frame-wikihow-task.json"; //original data point file
		List<WikiHowTaskFrame> allframe = InformationExtraction.getAllFrame(input);
		int total = 0;
		for (int i = 0; i < results.size(); i++) {
			System.out.println("Cluster " + i + ": " + results.get(i).getSuperClusterMembers().size());
			Writer out = new BufferedWriter(new OutputStreamWriter(
					new FileOutputStream(output+i+".json"), "utf-8"));
			 List<Integer> actitiviesID =
					 results.get(i).getSuperClusterMembers();
			 for (int j=0; j<allframe.size(); j++){
				 if (actitiviesID.contains(allframe.get(j).getID())){
					 out.write(allframe.get(j).toJsonObject().toJSONString() + "\n");
					 allframe.remove(j);
					 j--;
					 total++;
				 }
			 }
			 
			 out.close();
		}
		
		long endTime   = System.currentTimeMillis();
		long totalTime = (endTime - startTime)/60000;
		System.out.println("Number of activity frames: " + total);
		System.out.println("Running Time: " + totalTime + " minute(s)");
	}

}