protected ClusterNode mergeClusters(ClusterNode first, ClusterNode second) {
for (Node node : second.getPrimitives()) {
first.add(node);
graph.setNodeAliasMapping(node.getIndex(), first);
}
graph.removeClusterNode(second);
return first;
}
/*
* (non-Javadoc)
*
* @see
* org.processmining.mining.fuzzymining.graph.transform.FuzzyGraphTransformer
* #transform(org.processmining.mining.fuzzymining.graph.FuzzyGraph)
*/
public void transform(MutableFuzzyGraph graph) {
this.graph = graph;
// clean up edges
cleanUpEdges(threshold * 0.1, 1.0); // threshold * 0.3);
// determine simplification victims
ArrayList<Node> simplificationVictims = new ArrayList<Node>();
Node n;
for (int i = 0; i < graph.getNumberOfInitialNodes(); i++) {
n = graph.getPrimitiveNode(i);
if (n.getSignificance() < threshold) {
simplificationVictims.add(n);
}
}
// create clusters
int clusterIndex = graph.getNumberOfInitialNodes() + 1;
while (simplificationVictims.size() > 0) {
ClusterNode cluster = new ClusterNode(graph, clusterIndex);
graph.addClusterNode(cluster);
clusterIndex++;
boolean nodeAdded = true;
while (nodeAdded == true) {
nodeAdded = false;
ArrayList<Node> clustered = new ArrayList<Node>();
for (Node node : simplificationVictims) {
if (shouldMergeWith(node, cluster) == true) {
cluster.add(node);
graph.setNodeAliasMapping(node.getIndex(), cluster);
clustered.add(node);
nodeAdded = true;
}
}
simplificationVictims.removeAll(clustered);
}
}
// merge clusters
mergeAllClusters();
// remove unary clusters
for (int i = graph.getClusterNodes().size() - 1; i >= 0; i--) {
ClusterNode cluster = graph.getClusterNodes().get(i);
if (cluster.getPrimitives().size() == 1) {
// unary cluster; remove
for (int k = 0; k < graph.getNumberOfInitialNodes(); k++) {
Node mapping = graph.getNodeMappedTo(k);
if (mapping != null && mapping.equals(cluster)) {
graph.setNodeAliasMapping(k, null);
}
}
graph.removeClusterNode(cluster);
}
}
// clean up edges
cleanUpEdges(threshold, threshold);
}
protected boolean shouldMergeWith(Node node, ClusterNode cluster) {
if (cluster.getPrimitives().size() == 0) {
return true; // always add first node
}
if (cluster.isDirectlyConnectedTo(node) == true && cluster.contains(node) == false) {
double ownSignificance = 0.0;
double otherSignificance = 0.0;
double ownCorrelation = 0.0;
double otherCorrelation = 0.0;
// aggregate correlation and significance of edges between the node
// in
// question and connected nodes, separately for edges to/from
// cluster and
// to/from other nodes.
for (int i = 0; i < graph.getNumberOfInitialNodes(); i++) {
if (i == node.getIndex()) {
continue; // ignore self-references
}
if (cluster.contains(graph.getPrimitiveNode(i))) {
ownSignificance += graph.getBinarySignificance(node.getIndex(), i);
ownSignificance += graph.getBinarySignificance(i, node.getIndex());
ownCorrelation += graph.getBinaryCorrelation(node.getIndex(), i);
ownCorrelation += graph.getBinaryCorrelation(i, node.getIndex());
} else {
otherSignificance += graph.getBinarySignificance(node.getIndex(), i);
otherSignificance += graph.getBinarySignificance(i, node.getIndex());
otherCorrelation += graph.getBinaryCorrelation(node.getIndex(), i);
otherCorrelation += graph.getBinaryCorrelation(i, node.getIndex());
}
}
// make a decision
return (ownCorrelation > otherCorrelation || ownSignificance > otherSignificance);
} else {
// no connection - no merge.
return false;
}
}