@SuppressWarnings("unchecked")
@Override
public List<String> getClusterByCarrotVersion2(String query) {
// TODO Auto-generated method stub
List<String> strs = new ArrayList<String>();
final Controller controller = ControllerFactory.createPooling();
final Map<String, Object> luceneGlobalAttributes = new HashMap<String, Object>();
LuceneDocumentSourceDescriptor.attributeBuilder(luceneGlobalAttributes).directory(directory);
SimpleFieldMapperDescriptor.attributeBuilder(luceneGlobalAttributes)
.titleField(TITLE_FIELD)
.contentField(CONTENTS_FIELD)
.searchFields(Arrays.asList(new String[] {TITLE_FIELD, CONTENTS_FIELD}));
controller.init(
new HashMap<String, Object>(),
new ProcessingComponentConfiguration(
LuceneDocumentSource.class, "lucene", luceneGlobalAttributes));
final Map<String, Object> processingAttributes = Maps.newHashMap();
CommonAttributesDescriptor.attributeBuilder(processingAttributes).query(query);
ProcessingResult process =
controller.process(
processingAttributes, "lucene", LingoClusteringAlgorithm.class.getName());
for (Cluster c : process.getClusters()) {
strs.add(c.getLabel() + " >>>> " + c.getAllDocuments().size());
}
return strs;
}
int calculateH(Cluster cluster) {
final Map<Object, Integer> documentCountByPartition =
getDocumentCountByPartition(cluster.getAllDocuments());
final ArrayList<Integer> counts = Lists.newArrayList();
counts.addAll(documentCountByPartition.values());
return calculateH(counts);
}
@SuppressWarnings("unchecked")
double calculate(Cluster cluster, int partitionCount) {
int clusterPartitionAssignments = 0;
for (Document document : cluster.getAllDocuments()) {
clusterPartitionAssignments +=
((Collection<Object>) document.getField(Document.PARTITIONS)).size();
}
final double worstCaseH = calculateWorstCaseH(clusterPartitionAssignments, partitionCount);
if (worstCaseH == 0) {
return 0;
} else {
return calculateH(cluster) / worstCaseH;
}
}
