From 1cfee42c4792b9a2102fd6478ac9ac928b9c543b Mon Sep 17 00:00:00 2001
From: Francesco <francesco.bariatti@irisa.fr>
Date: Fri, 8 Apr 2022 11:47:57 +0200
Subject: [PATCH] Prepare: getting named entities of different types, without
 duplicate names

---
 prepare/src/main/Main.scala           | 19 +++++++++++++------
 prepare/src/main/NamedEntity.scala    |  4 ++--
 prepare/src/main/NamedEntitySet.scala | 16 ++++++++++++++++
 3 files changed, 31 insertions(+), 8 deletions(-)
 create mode 100644 prepare/src/main/NamedEntitySet.scala

diff --git a/prepare/src/main/Main.scala b/prepare/src/main/Main.scala
index a8ba710..f5ab08d 100644
--- a/prepare/src/main/Main.scala
+++ b/prepare/src/main/Main.scala
@@ -5,7 +5,6 @@ import org.apache.jena.reasoner.ReasonerRegistry
 import java.io.InputStream
 
 object Main extends App {
-
 	{
 		println("Loading knowledge graph...")
 		val kg: InfModel = ModelFactory.createInfModel(ReasonerRegistry.getTransitiveReasoner, ModelFactory.createDefaultModel)
@@ -17,8 +16,12 @@ object Main extends App {
 			})
 		kg.read(mondialDataset, null, "TTL");
 		printf("Loaded %s triples\n", kg.listStatements().toList.size)
-		val countries = getMondialNamedEntities(kg, "Country")
-		printf("There are %d countries in the dataset\n", countries.size)
+		val guessableEntities = NamedEntitySet() ++
+			getMondialNamedEntities(kg, "Country") ++
+			getMondialNamedEntities(kg, "City") ++
+			getMondialNamedEntities(kg, "Continent") ++
+			getMondialNamedEntities(kg, "Sea")
+		printf("Found %d guessable named entities\n", guessableEntities.set.size)
 	}
 
 	def getMondialNamedEntities(kg: Model, prefixedType: String): List[NamedEntity] = {
@@ -27,16 +30,20 @@ object Main extends App {
 		val MONDIAL_PREFIX = "http://www.semwebtech.org/mondial/10/meta#"
 		val fullType = MONDIAL_PREFIX + prefixedType;
 		val nameProperty = MONDIAL_PREFIX + "name"
-		val queryString = String.format("SELECT ?entity ?name WHERE { ?entity a <%s> ; <%s> ?name. }", fullType, nameProperty)
+		val queryString = s"SELECT ?entity ?name WHERE { ?entity a <${fullType}> ; <${nameProperty}> ?name. }"
 		val resultSet = QueryExecutionFactory.create(queryString, kg).execSelect()
 		while (resultSet.hasNext) {
 			val solution = resultSet.next()
 			result = result :+ new NamedEntity(
 				solution.getLiteral("?name").getString,
-				solution.getResource("?entity")
+				solution.getResource("?entity"),
+				prefixedType
 			)
 		}
+		if (result.isEmpty) {
+			System.err.printf("Error: found no named entities of type %s\n", prefixedType)
+			System.exit(1)
+		}
 		result
 	}
-
 }
diff --git a/prepare/src/main/NamedEntity.scala b/prepare/src/main/NamedEntity.scala
index dbdb7cc..91937c4 100644
--- a/prepare/src/main/NamedEntity.scala
+++ b/prepare/src/main/NamedEntity.scala
@@ -1,5 +1,5 @@
 import org.apache.jena.rdf.model.Resource
 
-class NamedEntity(val Name: String, val entity: Resource) {
-
+class NamedEntity(val Name: String, val entity: Resource, val readableType: String) {
 }
+
diff --git a/prepare/src/main/NamedEntitySet.scala b/prepare/src/main/NamedEntitySet.scala
new file mode 100644
index 0000000..f4bd03c
--- /dev/null
+++ b/prepare/src/main/NamedEntitySet.scala
@@ -0,0 +1,16 @@
+class NamedEntitySet(val set: Set[NamedEntity] = Set()) {
+
+	def ++(iterable: Iterable[NamedEntity]): NamedEntitySet = {
+		val nameSet = set.map(e => e.Name)
+		val duplicateNameEntity = iterable.find(e => nameSet.contains(e.Name))
+		if (duplicateNameEntity.nonEmpty) {
+			System.err.printf("Error: multiple entities exist with name %s\n", duplicateNameEntity.get.Name)
+			System.exit(1)
+		}
+		NamedEntitySet(set ++ iterable)
+	}
+}
+
+object NamedEntitySet {
+	def apply(set: Set[NamedEntity] = Set()): NamedEntitySet = new NamedEntitySet(set)
+}