For dbpedia data, prioritise manual conflict resolutions

2 files changed, 22 insertions, 21 deletions

diff --git a/backend/data/dbpPickedLabels.txt b/backend/data/dbpPickedLabels.txt
index d8f939e..1fa1e71 100644
--- a/backend/data/dbpPickedLabels.txt
+++ b/backend/data/dbpPickedLabels.txt
@@ -609,6 +609,7 @@ Tawera
 Telphusa
 Termitomyces schimperi
 Theba
+Theria
 Thestor
 Thomomys
 Thria
diff --git a/backend/data/genDbpData.py b/backend/data/genDbpData.py
index 1dce001..0ba1ef1 100755
--- a/backend/data/genDbpData.py
+++ b/backend/data/genDbpData.py
@@ -67,6 +67,27 @@ for name in nodeToLabel:
 	del nameToVariants[name]
 nodeToLabel["cellular organisms"] = "organism" # Special case for root node
 print("Number of conflicts: {}".format(len(nameToVariants)))
+# Try conflict resolution via picked-labels
+print("Resolving conflicts using picked-labels")
+with open(pickedLabelsFile) as file:
+	for line in file:
+		pickedLabel = line.rstrip()
+		name = pickedLabel.lower()
+		if name in nameToVariants:
+			nodeToLabel[name] = pickedLabel
+			del nameToVariants[name]
+		else:
+			match = nameVariantRegex.match(pickedLabel)
+			if match == None:
+				print("WARNING: Picked label {} not found (1)".format(pickedLabel), file=sys.stderr)
+			else:
+				name = match.group(1)
+				if name not in nameToVariants:
+					print("WARNING: Picked label {} not found (2)".format(pickedLabel), file=sys.stderr)
+				else:
+					nodeToLabel[name] = pickedLabel
+					del nameToVariants[name]
+print("Number of conflicts: {}".format(len(nameToVariants)))
 # Try conflict resolution via category-list
 	# Does a generic-category pass first (avoid stuff like Pan being classified as a horse instead of an ape)
 print("Resolving conflicts using category-list")
@@ -164,27 +185,6 @@ for (label, type) in dbpCur.execute("SELECT label, type from labels INNER JOIN t
 					nodeToLabel[name] = label
 					del nameToVariants[name]
 print("Number of conflicts: {}".format(len(nameToVariants)))
-# Try conflict resolution via picked-labels
-print("Resolving conflicts using picked-labels")
-with open(pickedLabelsFile) as file:
-	for line in file:
-		pickedLabel = line.rstrip()
-		name = pickedLabel.lower()
-		if name in nameToVariants:
-			nodeToLabel[name] = pickedLabel
-			del nameToVariants[name]
-		else:
-			match = nameVariantRegex.match(pickedLabel)
-			if match == None:
-				print("WARNING: Picked label {} not found (1)".format(pickedLabel), file=sys.stderr)
-			else:
-				name = match.group(1)
-				if name not in nameToVariants:
-					print("WARNING: Picked label {} not found (2)".format(pickedLabel), file=sys.stderr)
-				else:
-					nodeToLabel[name] = pickedLabel
-					del nameToVariants[name]
-print("Number of conflicts: {}".format(len(nameToVariants)))
 # Associate nodes with IRIs
 print("Getting nodes IRIs")
 nodeToIri = {}