private String getEdgeString(FeatureTree tree, TreeNode parentNode) {
StringBuilder sb = new StringBuilder();
for (TreeNode child : parentNode.getChildren()) {
if (child.getCost() != 0) {
sb.append(getNodeTypeAndDatabaseIdString(tree, parentNode));
sb.append(getNodeTypeAndDatabaseIdString(tree, child));
String value = String.valueOf(child.getValueOfCandidateFeature());
value = fixPlusAndMinusSigns(value);
EdgeLabels.FEATURE_VALUE.setAttributeValue(value);
sb.append(EdgeLabels.FEATURE_VALUE.getDefaultValueString());
sb.append(EdgeLabels.FEATURE_VALUE.toString());
value = String.valueOf(parentNode.getAppliedCandidate());
value = fixPlusAndMinusSigns(value);
EdgeLabels.CANDIDATE_USED.setAttributeValue(value);
sb.append(EdgeLabels.CANDIDATE_USED.toString());
sb.append("\n");
}
}
return sb.toString();
}
private String getNodeTypeAndDatabaseIdString(FeatureTree tree, TreeNode node) {
StringBuilder sb = new StringBuilder();
// type-attribute
// level of the tree
sb.append(tree.getBabyTreeType());
sb.append("/");
if (tree.getClass() == JointlyCodingFeatureTree.class) {
sb.append(tree.getAlignmentType());
sb.append("/");
}
// name of the feature
sb.append(tree.getFeatureName());
sb.append("_");
// unique node number - database_id -attribute
int nodeIdentity = node.getNodeIdentityNumber();
sb.append(nodeIdentity);
sb.append(" ");
return sb.toString();
}
private void printTree(FeatureTree tree) {
minEntropy = Double.POSITIVE_INFINITY;
maxEntropy = Double.NEGATIVE_INFINITY;
for (TreeNode node : tree.getNodes()) {
// if node matrix contains only zeros, don't print
if (node.getCost() != 0) {
node.setNodeIdentityNumber(nodeCounter);
nodeCounter += 1;
double nodeEntropy = computeEntropy(node.getCountMatrix());
if (nodeEntropy > maxEntropy) {
maxEntropy = nodeEntropy;
}
if (nodeEntropy < minEntropy) {
minEntropy = nodeEntropy;
}
}
}
for (TreeNode node : tree.getNodes()) {
try {
if (node.getCost() != 0) {
fw.write(getEdgeString(tree, node));
fw.write(getNodeString(tree, node));
}
} catch (IOException ex) {
Logger.getLogger(TreeGraphPrinter.class.getName()).log(Level.SEVERE, null, ex);
}
}
}
private String getMatrixString(FeatureTree tree, TreeNode node) {
StringBuilder sb = new StringBuilder();
int[][] matrix = node.getCountMatrix();
if (tree.getClass() != JointlyCodingFeatureTree.class) {
return Arrays.toString(matrix[0]);
} else {
int counter = 0;
sb.append(" ");
for (char c : tree.getColumnLabels()) {
sb.append(c).append(" ");
}
sb.append("\\n");
for (int[] row : matrix) {
sb.append(tree.getRowLabels().get(counter)).append(" ");
sb.append(Arrays.toString(row)).append("\\n");
counter++;
}
// sb.deleteCharAt(sb.length()-1);
}
return sb.toString();
}
private String getNodeString(FeatureTree tree, TreeNode parentNode) {
StringBuilder sb = new StringBuilder();
// each node line begins with this
sb.append("# _attributes ");
String language = tree.getLanguage();
// the default name of the node; here the name of the feature
sb.append(getNodeTypeAndDatabaseIdString(tree, parentNode));
if (parentNode.isRootNode()) {
// add some extra attributes for root node
sb.append(getAttributeString(NodeLabels.TOTAL_COST_OF_TREE, tree.getTotalTreeCost()));
sb.append(getAttributeString(NodeLabels.MODEL_COST_OF_TREE, tree.getModelCost()));
sb.append(getAttributeString(NodeLabels.DATA_COST_OF_TREE, tree.getDataCost()));
sb.append(getAttributeString(NodeLabels.FEATURE_NAME, tree.getFeatureName()));
sb.append(
getAttributeString(
NodeLabels.TREE_LEVEL, tree.getBabyTreeType().toString() + ":" + language));
sb.append(getAttributeString(NodeLabels.TREE_TYPE, tree.getTreeType()));
sb.append(getAttributeString(NodeLabels.COLUMN_LABELS, tree.getColumnLabels()));
if (tree.getClass() == JointlyCodingFeatureTree.class) {
sb.append(getAttributeString(NodeLabels.ROW_LABELS, tree.getRowLabels()));
sb.append(getAttributeString(NodeLabels.ALIGNMENT_TYPE, tree.getAlignmentType()));
}
if (tree.getBabyTreeType() != BabyTreeType.JOINT) {
sb.append(getAttributeString(NodeLabels.POS, getRootNodePositionString(tree, parentNode)));
}
sb.append("queryset=root ");
} else {
// not root node
sb.append(
getAttributeString(
NodeLabels.FILL,
getColor(minEntropy, maxEntropy, computeEntropy(parentNode.getCountMatrix()))));
}
// default label, if not defined, the getNodeDatabaseIdString is shown in nodes instead
sb.append(getAttributeString(NodeLabels.LABEL, getMatrixString(tree, parentNode)));
// attributes that all nodes share
sb.append(getAttributeString(NodeLabels.MATRIX_COST, parentNode.getCost()));
sb.append(getAttributeString(NodeLabels.MATRIX, getMatrixString(tree, parentNode)));
sb.append(
getAttributeString(
NodeLabels.FEATURE_VALUE_IN_PREVIOUS_SPLIT, parentNode.getValueOfCandidateFeature()));
sb.append(
getAttributeString(NodeLabels.NUMBER_OF_CANDIDATES, parentNode.getCandidates().size()));
sb.append(getAttributeString(NodeLabels.ENTROPY, computeEntropy(parentNode.getCountMatrix())));
// inner nodes have been split
if (parentNode.getAppliedCandidate() != null) {
sb.append(getAttributeString(NodeLabels.BEST_CANDIDATE, parentNode.getAppliedCandidate()));
sb.append(
getAttributeString(
NodeLabels.BEST_CANDIDATE_COST, parentNode.getAppliedCandidate().getCost()));
}
sb.append("\n");
return sb.toString();
}