private static Map<Long, Map<Boolean, Map<Integer, Pair<SingleDTGraph, List<Double>>>>>
createDataSetCache(
RDFDataSet tripleStore,
LargeClassificationDataSet data,
long[] seeds,
double fraction,
int minSize,
int maxClasses,
int[] depths,
boolean[] inference) {
Map<Long, Map<Boolean, Map<Integer, Pair<SingleDTGraph, List<Double>>>>> cache =
new HashMap<Long, Map<Boolean, Map<Integer, Pair<SingleDTGraph, List<Double>>>>>();
for (long seed : seeds) {
cache.put(seed, new HashMap<Boolean, Map<Integer, Pair<SingleDTGraph, List<Double>>>>());
data.createSubSet(seed, fraction, minSize, maxClasses);
for (boolean inf : inference) {
cache.get(seed).put(inf, new HashMap<Integer, Pair<SingleDTGraph, List<Double>>>());
for (int depth : depths) {
System.out.println("Getting Statements...");
Set<Statement> stmts =
RDFUtils.getStatements4Depth(
tripleStore, data.getRDFData().getInstances(), depth, inf);
System.out.println("# Statements: " + stmts.size());
stmts.removeAll(new HashSet<Statement>(data.getRDFData().getBlackList()));
System.out.println("# Statements: " + stmts.size() + ", after blackList");
System.out.println("Building Graph...");
SingleDTGraph graph =
RDFUtils.statements2Graph(
stmts, RDFUtils.REGULAR_LITERALS, data.getRDFData().getInstances(), true);
System.out.println(
"Built Graph with "
+ graph.getGraph().nodes().size()
+ ", and "
+ graph.getGraph().links().size()
+ " links");
cache
.get(seed)
.get(inf)
.put(
depth,
new Pair<SingleDTGraph, List<Double>>(
graph, new ArrayList<Double>(data.getTarget())));
}
}
}
return cache;
}
private static void computeGraphStatistics(
RDFDataSet tripleStore, ClassificationDataSet ds, boolean[] inference, int[] depths) {
Map<Boolean, Map<Integer, Pair<Double, Double>>> stats =
new HashMap<Boolean, Map<Integer, Pair<Double, Double>>>();
for (boolean inf : inference) {
stats.put(inf, new HashMap<Integer, Pair<Double, Double>>());
for (int depth : depths) {
Set<Statement> st =
RDFUtils.getStatements4Depth(tripleStore, ds.getRDFData().getInstances(), depth, inf);
st.removeAll(ds.getRDFData().getBlackList());
DTGraph<String, String> graph = RDFUtils.statements2Graph(st, RDFUtils.REGULAR_LITERALS);
List<DTNode<String, String>> instanceNodes =
RDFUtils.findInstances(graph, ds.getRDFData().getInstances());
graph = RDFUtils.simplifyInstanceNodeLabels(graph, instanceNodes);
GraphList<DTGraph<String, String>> graphs =
RDFUtils.getSubGraphs(graph, instanceNodes, depth);
double v = 0;
double e = 0;
for (DTGraph<String, String> g : graphs.getGraphs()) {
v += g.nodes().size();
e += g.links().size();
}
v /= graphs.numInstances();
e /= graphs.numInstances();
stats.get(inf).put(depth, new Pair<Double, Double>(v, e));
}
}
for (boolean k1 : stats.keySet()) {
System.out.println("Inference: " + k1);
for (int k2 : stats.get(k1).keySet()) {
System.out.println(
"Depth "
+ k2
+ ", vertices: "
+ (stats.get(k1).get(k2).getFirst())
+ " , edges: "
+ (stats.get(k1).get(k2).getSecond()));
}
}
}
private void init(RDFDataSet dataset, List<Resource> instances, List<Statement> blackList) {
Set<Statement> stmts = RDFUtils.getStatements4Depth(dataset, instances, depth, inference);
stmts.removeAll(blackList);
graph = RDFUtils.statements2Graph(stmts, RDFUtils.REGULAR_LITERALS, instances, true);
}