public void computePageRanks(CategoryGraph graph) {
LOG.info("computing category page ranks...");
// initialize page rank
long sumCredits = graph.catPages.length; // each category gets 1 credit to start
for (int i = 0; i < graph.catPages.length; i++) {
sumCredits += graph.catPages[i].length; // one more credit per page that references it.
}
for (int i = 0; i < graph.catPages.length; i++) {
graph.catCosts[i] = (1.0 + graph.catPages[i].length) / sumCredits;
}
for (int i = 0; i < 20; i++) {
LOG.info("performing page ranks iteration {0}.", i);
double error = onePageRankIteration(graph);
LOG.info("Error for iteration is {0}.", error);
if (error == 0) {
break;
}
}
Integer sortedIndexes[] = new Integer[graph.catCosts.length];
for (int i = 0; i < graph.catParents.length; i++) {
graph.catCosts[i] = 1.0 / -Math.log(graph.catCosts[i]);
sortedIndexes[i] = i;
}
LOG.info("finished computing page ranks...");
final double[] costs = graph.catCosts;
Arrays.sort(
sortedIndexes,
new Comparator<Integer>() {
@Override
public int compare(Integer i1, Integer i2) {
Double pr1 = costs[i1];
Double pr2 = costs[i2];
return -1 * pr1.compareTo(pr2);
}
});
StringBuilder b = new StringBuilder();
for (int i = 0; i < 20; i++) {
int j = sortedIndexes[i];
b.append("" + i + ". " + graph.cats[j] + "=" + graph.catCosts[j]);
b.append(", ");
}
graph.minCost = graph.catCosts[sortedIndexes[sortedIndexes.length - 1]];
LOG.info("Min cat cost: " + graph.minCost);
LOG.info("Top cat costs: " + b.toString());
}
public double onePageRankIteration(CategoryGraph graph) {
double nextRanks[] = new double[graph.catCosts.length];
Arrays.fill(nextRanks, (1.0 - DAMPING_FACTOR) / graph.catCosts.length);
for (int i = 0; i < graph.catParents.length; i++) {
int d = graph.catParents[i].length; // degree
double pr = graph.catCosts[i]; // current page-rank
for (int j : graph.catParents[i]) {
nextRanks[j] += DAMPING_FACTOR * pr / d;
}
}
double diff = 0.0;
for (int i = 0; i < graph.catParents.length; i++) {
diff += Math.abs(graph.catCosts[i] - nextRanks[i]);
}
graph.catCosts = nextRanks;
return diff;
}