Add French NER

grobid-ner/pom.xml

@@ -6,13 +6,13 @@
 	<parent>
 		<groupId>org.grobid</groupId>
 		<artifactId>grobid-parent</artifactId>
-		<version>0.4.1-SNAPSHOT</version>
+		<version>0.4.2-SNAPSHOT</version>
 		<relativePath>../../pom.xml</relativePath>
 	</parent>
 
 	<groupId>org.grobid</groupId>
 	<artifactId>grobid-ner</artifactId>
-	<version>0.4.1-SNAPSHOT</version>
+	<version>0.4.2-SNAPSHOT</version>
 	<packaging>jar</packaging>
 	<name>${project.artifactId}</name>
 
@@ -44,12 +44,12 @@
 		<dependency>
 			<groupId>org.grobid</groupId>
 			<artifactId>grobid-core</artifactId>
-			<version>0.4.1-SNAPSHOT</version>
+			<version>0.4.2-SNAPSHOT</version>
 		</dependency>
 		<dependency>
 			<groupId>org.grobid</groupId>
 			<artifactId>grobid-trainer</artifactId>
-			<version>0.4.1-SNAPSHOT</version>
+			<version>0.4.2-SNAPSHOT</version>
 		</dependency>
 
 		<!--dependency>
@@ -93,7 +93,7 @@
 		<dependency>
 			<groupId>org.apache.commons</groupId>
 			<artifactId>commons-lang3</artifactId>
-			<version>3.0.1</version>
+			<version>3.4</version>
 		</dependency>
 		<dependency>
 			<groupId>org.apache.commons</groupId>
@@ -239,6 +239,29 @@
 				</plugins>
 			</build>
 		</profile>
+		<profile>
+			<id>train_nerfr</id>
+			<build>
+				<plugins>
+					<plugin>
+						<groupId>org.codehaus.mojo</groupId>
+						<artifactId>exec-maven-plugin</artifactId>
+						<version>1.1.1</version>
+						<executions>
+							<execution>
+								<phase>generate-resources</phase>
+								<goals>
+									<goal>java</goal>
+								</goals>
+								<configuration>
+									<mainClass>org.grobid.trainer.NERFrenchTrainer</mainClass>
+								</configuration>
+							</execution>
+						</executions>
+					</plugin>
+				</plugins>
+			</build>
+		</profile>
 		<!-- mvn generate-resources -Ptrain_nersense -e -->
 		<profile>
 			<id>train_nersense</id>

grobid-ner/resources/dataset/nerfr/crfpp-templates/nerfr.template

@@ -0,0 +1,108 @@
+# Token 0
+U00:%x[-4,0]
+U01:%x[-3,0]
+U02:%x[-2,0]
+U03:%x[-1,0]
+U04:%x[0,0]
+U05:%x[1,0]
+U06:%x[2,0]
+U07:%x[3,0]
+U08:%x[4,0]
+#U09:%x[-1,0]/%x[0,0]
+#U0A:%x[0,0]/%x[1,0]
+#U0B:%x[1,0]/%x[2,0]
+#U0C:%x[-2,0]/%x[-1,0]
+
+# Lowercase token 1
+U10:%x[-2,1]
+U11:%x[-1,1]
+U12:%x[0,1]
+U13:%x[1,1]
+U14:%x[2,1]
+
+# Prefix 1-4 characters (2-6)
+U20:%x[0,2]
+U21:%x[0,3]
+U22:%x[0,4]
+U23:%x[0,5]
+U24:%x[0,6]
+
+# Suffix 1-4 characters (7-11)
+U30:%x[0,7]
+U31:%x[0,8]
+U32:%x[0,9]
+U33:%x[0,10]
+U34:%x[0,11]
+
+# Capitalization (12)
+U40:%x[0,12]
+U41:%x[1,12]
+U42:%x[-1,12]
+U43:%x[-2,12]
+#U44:%x[0,12]/%x[1,12]
+#U45:%x[-1,12]/%x[0,12]
+
+# Digits (13)
+U50:%x[0,13]
+U51:%x[-1,13]
+U52:%x[1,13]
+
+# Lexical information (14-20)
+U80:%x[0,14]
+U81:%x[0,15]
+U82:%x[0,16]
+U83:%x[0,17]
+U84:%x[0,18]
+U85:%x[0,19]
+U86:%x[0,20]
+U89:%x[-1,14]
+U8A:%x[-1,15]
+U8B:%x[-1,16]
+U8C:%x[-1,17]
+U8D:%x[-1,18]
+U8E:%x[-1,19]
+U8F:%x[-1,20]
+U8I:%x[1,14]
+U8J:%x[1,15]
+U8K:%x[1,16]
+U8L:%x[1,17]
+U8M:%x[1,18]
+U8N:%x[1,19]
+U8O:%x[1,20]
+
+# lexical feature: belongs to a known location (21)
+U90:%x[0,21]
+U91:%x[-1,21]
+U92:%x[1,21]
+U93:%x[-2,21]
+
+# lexical feature: belongs to a known person title (22)
+UA0:%x[0,22]
+UA1:%x[-1,22]
+UA2:%x[1,22]
+UA3:%x[-2,22]
+
+# lexical feature: belongs to a known organisation (23)
+UB0:%x[0,23]
+UB1:%x[-1,23]
+UB2:%x[1,23]
+UB3:%x[-2,23]
+
+# lexical feature: belongs to a known organisation form name (24)
+UC0:%x[0,24]
+UC1:%x[-1,24]
+UC2:%x[1,24]
+UC3:%x[-2,24]
+
+# word shape (25)
+UD0:%x[0,25]
+UD1:%x[-1,25]
+UD2:%x[1,25]
+
+# word shape trimmed (26)
+UE0:%x[0,26]
+UE1:%x[-1,26]
+UE2:%x[1,26]
+
+# Output
+B

grobid-ner/src/main/java/org/grobid/core/data/Entity.java

@@ -14,6 +14,22 @@
  *
  */
 public class Entity implements Comparable<Entity> {   
+
+	// Orign of the entity definition
+	public enum Origin {
+		GROBID	("grobid"),
+		USER	("user");
+
+		private String name;
+
+		private Origin(String name) {
+          	this.name = name;
+		}
+
+		public String getName() {
+			return name;
+		}
+	};
 
 	// name of the entity = entity type                   
 	private String rawName = null;
@@ -32,7 +48,7 @@ public class Entity implements Comparable<Entity> {
 
 	// probability of the entity in context, if defined
 	private double prob = 1.0;
-	
+
 	// confidence score of the entity in context, if defined
 	private double conf = 0.8;
 
@@ -43,9 +59,7 @@ public class Entity implements Comparable<Entity> {
 	private BoundingBox box = null;
 
 	// orign of the entity definition
-	public static int GROBID = 0;
-	public static int USER = 1;
-	private int origin = 0;
+	private Origin origin = Origin.GROBID;
 
     public Entity() {
 		this.offsets = new OffsetPosition();
@@ -160,11 +174,11 @@ public void setSense(Sense sense) {
 		this.sense = sense;
 	}
 
-	public int getOrigin() {
+	public Origin getOrigin() {
 		return origin;
 	}
 
-	public void setOrigin(int origin) {
+	public void setOrigin(Origin origin) {
 		this.origin = origin;
 	}