public static HashMap<Integer, List<Integer>> groupSimilarUMLSCons(
EncodedEntityStructure umlsGroup,
Set<GenericProperty> preRanAtts,
int size,
FormRepository gi) {
EncodedAnnotationMapping set = new EncodedAnnotationMapping();
Set<Integer> srcs = new HashSet<Integer>();
srcs.add(umlsGroup.getStructureId());
// Int2FloatMap idfMap = TFIDFTokenWeightGenerator.getInstance().generateIDFValues(srcs, size);
// Map<String,Object> externalObjects = new HashMap<String,Object>();
// externalObjects.put(TFIDFMatcher.IDF_MAP_SOURCE, idfMap);
// externalObjects.put(TFIDFMatcher.IDF_MAP_TARGET, idfMap);
// externalObjects.put(TFIDFMatcher.TFIDF_SOURCE_SEPARATED,false);
long time = System.currentTimeMillis();
MatchOperator mop =
new MatchOperator(
RegisteredMatcher.TRIGRAM_MATCHER,
AggregationFunction.MAX,
preRanAtts,
preRanAtts,
0.35f);
// mop.setGlobalObjects(externalObjects);
ExecutionTree tree = new ExecutionTree();
tree.addOperator(mop);
try {
set = gi.getMatchManager().matchEncoded(umlsGroup, umlsGroup, tree, null);
log.debug("matched group");
} catch (MatchingExecutionException e1) {
// TODO Auto-generated catch block
e1.printStackTrace();
}
HashMap<Integer, Node> nodeMap = new HashMap<Integer, Node>();
boolean isChange = true;
int counter = 0;
HashSet<EntityAnnotation> toRemoveCorrs = new HashSet<EntityAnnotation>();
for (EntityAnnotation c : set.getAnnotations()) {
if (c.getSrcId() == c.getTargetId()) {
toRemoveCorrs.add(c);
}
}
for (EntityAnnotation c : toRemoveCorrs) {
set.removeAnnotation(c.getSrcId(), c.getTargetId());
}
while (isChange && counter < 10000) {
counter++;
if (set.getNumberOfAnnotations() != 0) {
for (EntityAnnotation cor : set.getAnnotations()) {
if (cor.getSrcId() != cor.getTargetId()) {
Node n = nodeMap.get(cor.getSrcId());
if (n == null) {
n = new Node();
n.ownId = cor.getSrcId();
n.minId = n.ownId;
nodeMap.put(n.ownId, n);
}
Node n2 = nodeMap.get(cor.getTargetId());
if (n2 == null) {
n2 = new Node();
n2.ownId = cor.getTargetId();
n2.minId = cor.getTargetId();
nodeMap.put(n2.ownId, n2);
}
int min = Math.min(n.minId, n2.minId);
if (n2.minId == min && n.minId == min) {
isChange = false;
} else {
isChange = true;
}
n.minId = min;
n2.minId = min;
}
}
} else {
for (int id : umlsGroup.getObjIds().keySet()) {
Node n = new Node();
n.ownId = id;
n.minId = n.ownId;
nodeMap.put(id, n);
}
isChange = false;
}
}
HashMap<Integer, List<Integer>> umlsGroups = new HashMap<Integer, List<Integer>>();
HashSet<Integer> notInGroup = new HashSet<Integer>();
for (int id : umlsGroup.getObjIds().keySet()) {
notInGroup.add(id);
}
notInGroup.removeAll(nodeMap.keySet());
for (Entry<Integer, Node> e : nodeMap.entrySet()) {
List<Integer> list = umlsGroups.get(e.getValue().minId);
if (list == null) {
list = new ArrayList<Integer>();
umlsGroups.put(e.getValue().minId, list);
}
list.add(e.getKey());
}
for (Integer nig : notInGroup) {
List<Integer> list = new ArrayList<Integer>();
umlsGroups.put(nig, list);
list.add(nig);
}
return umlsGroups;
}
public static HashMap<Integer, List<Integer>> groupSimilarUMLSConsByConnectedComponent(
EncodedEntityStructure umlsGroup,
Set<GenericProperty> preRanAtts,
int size,
FormRepository gi,
float threshold) {
EncodedAnnotationMapping set = new EncodedAnnotationMapping();
Set<Integer> srcs = new HashSet<Integer>();
srcs.add(umlsGroup.getStructureId());
MatchOperator mop =
new MatchOperator(
RegisteredMatcher.TRIGRAM_MATCHER,
AggregationFunction.MAX,
preRanAtts,
preRanAtts,
0.4f);
ExecutionTree tree = new ExecutionTree();
tree.addOperator(mop);
try {
log.debug(umlsGroup.getObjIds().size());
set = gi.getMatchManager().matchEncoded(umlsGroup, umlsGroup, tree, null);
log.debug("matched group");
} catch (MatchingExecutionException e1) {
// TODO Auto-generated catch block
e1.printStackTrace();
}
HashSet<EntityAnnotation> toRemoveCorrs = new HashSet<EntityAnnotation>();
HashMap<Integer, Node> nodeMap = new HashMap<Integer, Node>();
boolean isChange = true;
int counter = 0;
for (EntityAnnotation c : set.getAnnotations()) {
if (c.getSrcId() == c.getTargetId()) {
toRemoveCorrs.add(c);
}
}
for (EntityAnnotation c : toRemoveCorrs) {
set.removeAnnotation(c.getSrcId(), c.getTargetId());
}
while (isChange && counter < 10000) {
counter++;
if (set.getNumberOfAnnotations() != 0) {
for (EntityAnnotation cor : set.getAnnotations()) {
if (cor.getSrcId() != cor.getTargetId()) {
Node n = nodeMap.get(cor.getSrcId());
if (n == null) {
n = new Node();
n.ownId = cor.getSrcId();
n.minId = n.ownId;
nodeMap.put(n.ownId, n);
}
Node n2 = nodeMap.get(cor.getTargetId());
if (n2 == null) {
n2 = new Node();
n2.ownId = cor.getTargetId();
n2.minId = cor.getTargetId();
nodeMap.put(n2.ownId, n2);
}
int min = Math.min(n.minId, n2.minId);
if (n2.minId == min && n.minId == min) {
isChange = false;
} else {
isChange = true;
}
n.minId = min;
n2.minId = min;
}
}
} else {
for (int id : umlsGroup.getObjIds().keySet()) {
Node n = new Node();
n.ownId = id;
n.minId = n.ownId;
nodeMap.put(id, n);
}
isChange = false;
}
}
HashMap<Integer, List<Integer>> umlsGroups = new HashMap<Integer, List<Integer>>();
HashSet<Integer> notInGroup = new HashSet<Integer>();
for (int id : umlsGroup.getObjIds().keySet()) {
notInGroup.add(id);
}
notInGroup.removeAll(nodeMap.keySet());
for (Entry<Integer, Node> e : nodeMap.entrySet()) {
List<Integer> list = umlsGroups.get(e.getValue().minId);
if (list == null) {
list = new ArrayList<Integer>();
umlsGroups.put(e.getValue().minId, list);
}
list.add(e.getKey());
}
for (Integer nig : notInGroup) {
List<Integer> list = new ArrayList<Integer>();
umlsGroups.put(nig, list);
list.add(nig);
}
return umlsGroups;
}