public static Set<RepairedCell> getDuplicates(Set<RepairedCell> found) {
Set<RepairedCell> rst = new HashSet<RepairedCell>();
Map<String, List<RepairedCell>> tmpMap = new HashMap<String, List<RepairedCell>>();
for (RepairedCell cell : found) {
String key = cell.getRowId() + cell.getColumnId();
if (tmpMap.containsKey(key)) {
tmpMap.get(key).add(cell);
} else {
List<RepairedCell> tmpList = new ArrayList<RepairedCell>();
tmpList.add(cell);
tmpMap.put(key, tmpList);
}
}
if (tmpMap.size() > 0) {
for (String key : tmpMap.keySet()) {
if (tmpMap.get(key).size() > 1) {
for (RepairedCell cell : tmpMap.get(key)) {
rst.add(cell);
}
}
}
}
return rst;
}
public static Set<RepairedCell> getUnfound(Set<RepairedCell> truth, Set<RepairedCell> found) {
Set<RepairedCell> rst = new HashSet<RepairedCell>();
if (found.size() != 0) {
HashMap<Integer, HashSet<String>> foundMap = new HashMap<Integer, HashSet<String>>();
for (RepairedCell cell : found) {
HashSet<String> columnIds = null;
if (foundMap.get(cell.getRowId()) == null) {
columnIds = new HashSet<String>();
} else {
columnIds = foundMap.get(cell.getRowId());
}
columnIds.add(cell.getColumnId());
foundMap.put(cell.getRowId(), columnIds);
}
for (RepairedCell cell : truth) {
if (foundMap.get(cell.getRowId()) == null
|| !foundMap.get(cell.getRowId()).contains(cell.getColumnId())) {
rst.add(cell);
}
}
}
return rst;
}
public static double findAccuracy(Set<RepairedCell> truth, Set<RepairedCell> found) {
if (found.size() != 0) {
HashMap<Integer, HashSet<String>> foundMap = new HashMap<Integer, HashSet<String>>();
for (RepairedCell cell : found) {
HashSet<String> columnIds = null;
if (foundMap.get(cell.getRowId()) == null) {
columnIds = new HashSet<String>();
} else {
columnIds = foundMap.get(cell.getRowId());
}
columnIds.add(cell.getColumnId());
foundMap.put(cell.getRowId(), columnIds);
}
int tAndF = 0;
for (RepairedCell cell : truth) {
if (foundMap.get(cell.getRowId()) != null) {
if (foundMap.get(cell.getRowId()).contains(cell.getColumnId())) {
tAndF++;
}
}
}
double precision = tAndF * 1.0 / found.size(), recall = tAndF * 1.0 / truth.size();
if (debug) System.out.println("find precision = " + precision + ", find recall = " + recall);
return 2 * precision * recall / (precision + recall);
}
return 0;
}