List<String> strings = getStrings(args[1]);
double radius = Double.parseDouble(args[2]);
int blocking_size = Integer.parseInt(args[3]);
long vptree_start = System.currentTimeMillis();
Clusterer vptree_clusterer = new VPTreeClusterer(distance);
for (String s: strings) {
vptree_clusterer.populate(s);
}
List<Set<Serializable>> vptree_clusters = vptree_clusterer.getClusters(radius);
long vptree_elapsed = System.currentTimeMillis() - vptree_start;
int vptree_distances = distance.getCount();
distance.resetCounter();
long ngram_start = System.currentTimeMillis();
Clusterer ngram_clusterer = new NGramClusterer(distance,blocking_size);
for (String s: strings) {
ngram_clusterer.populate(s);
}
List<Set<Serializable>> ngram_clusters = ngram_clusterer.getClusters(radius);
long ngram_elapsed = System.currentTimeMillis() - ngram_start;
int ngram_distances = distance.getCount();
distance.resetCounter();
log("VPTree found " + vptree_clusters.size() + " in " + vptree_elapsed + " ms with " + vptree_distances + " distances\n");