Esempio n. 1
0
    @Override
    public void reduce(IntWritable nid, Iterable<PageRankNode> values, Context context)
        throws IOException, InterruptedException {
      int massMessages = 0;

      // Remember, PageRank mass is stored as a log prob.
      float mass = Float.NEGATIVE_INFINITY;
      for (PageRankNode n : values) {
        if (n.getType() == PageRankNode.Type.Structure) {
          // Simply pass along node structure.
          context.write(nid, n);
        } else {
          // Accumulate PageRank mass contributions.
          mass = sumLogProbs(mass, n.getPageRank());
          massMessages++;
        }
      }

      // Emit aggregated results.
      if (massMessages > 0) {
        intermediateMass.setNodeId(nid.get());
        intermediateMass.setType(PageRankNode.Type.Mass);
        intermediateMass.setPageRank(mass);

        context.write(nid, intermediateMass);
      }
    }
Esempio n. 2
0
    @Override
    public void reduce(IntWritable nid, Iterable<PageRankNode> iterable, Context context)
        throws IOException, InterruptedException {
      Iterator<PageRankNode> values = iterable.iterator();

      // Create the node structure that we're going to assemble back together from shuffled pieces.
      PageRankNode node = new PageRankNode();

      node.setType(PageRankNode.Type.Complete);
      node.setNodeId(nid.get());

      int massMessagesReceived = 0;
      int structureReceived = 0;

      float mass = Float.NEGATIVE_INFINITY;
      while (values.hasNext()) {
        PageRankNode n = values.next();

        if (n.getType().equals(PageRankNode.Type.Structure)) {
          // This is the structure; update accordingly.
          ArrayListOfIntsWritable list = n.getAdjacenyList();
          structureReceived++;

          node.setAdjacencyList(list);
        } else {
          // This is a message that contains PageRank mass; accumulate.
          mass = sumLogProbs(mass, n.getPageRank());
          massMessagesReceived++;
        }
      }

      // Update the final accumulated PageRank mass.
      node.setPageRank(mass);
      context.getCounter(PageRank.massMessagesReceived).increment(massMessagesReceived);

      // Error checking.
      if (structureReceived == 1) {
        // Everything checks out, emit final node structure with updated PageRank value.
        context.write(nid, node);

        // Keep track of total PageRank mass.
        totalMass = sumLogProbs(totalMass, mass);
      } else if (structureReceived == 0) {
        // We get into this situation if there exists an edge pointing to a node which has no
        // corresponding node structure (i.e., PageRank mass was passed to a non-existent node)...
        // log and count but move on.
        context.getCounter(PageRank.missingStructure).increment(1);
        LOG.warn(
            "No structure received for nodeid: " + nid.get() + " mass: " + massMessagesReceived);
        // It's important to note that we don't add the PageRank mass to total... if PageRank mass
        // was sent to a non-existent node, it should simply vanish.
      } else {
        // This shouldn't happen!
        throw new RuntimeException(
            "Multiple structure received for nodeid: "
                + nid.get()
                + " mass: "
                + massMessagesReceived
                + " struct: "
                + structureReceived);
      }
    }