[
https://issues.apache.org/jira/browse/OPENNLP-941?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16439514#comment-16439514
]
ASF GitHub Bot commented on OPENNLP-941:
----------------------------------------
kottmann closed pull request #308: [OPENNLP-941] Added eval support to
detokenizer
URL: https://github.com/apache/opennlp/pull/308
This is a PR merged from a forked repository.
As GitHub hides the original diff on merge, it is displayed below for
the sake of provenance:
As this is a foreign pull request (from a fork), the diff is supplied
below (as it won't show otherwise due to GitHub magic):
diff --git
a/opennlp-tools/src/main/java/opennlp/tools/cmdline/tokenizer/DetokenEvaluationErrorListener.java
b/opennlp-tools/src/main/java/opennlp/tools/cmdline/tokenizer/DetokenEvaluationErrorListener.java
new file mode 100644
index 000000000..4e313328c
--- /dev/null
+++
b/opennlp-tools/src/main/java/opennlp/tools/cmdline/tokenizer/DetokenEvaluationErrorListener.java
@@ -0,0 +1,53 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package opennlp.tools.cmdline.tokenizer;
+
+import java.io.OutputStream;
+
+import opennlp.tools.cmdline.EvaluationErrorPrinter;
+import opennlp.tools.tokenize.TokenSample;
+import opennlp.tools.tokenize.TokenizerEvaluationMonitor;
+import opennlp.tools.util.eval.EvaluationMonitor;
+
+/**
+ * A default implementation of {@link EvaluationMonitor} that prints
+ * to an output stream.
+ */
+public class DetokenEvaluationErrorListener extends
+ EvaluationErrorPrinter<TokenSample> implements TokenizerEvaluationMonitor {
+
+ /**
+ * Creates a listener that will print to System.err
+ */
+ public DetokenEvaluationErrorListener() {
+ super(System.err);
+ }
+
+ /**
+ * Creates a listener that will print to a given {@link OutputStream}
+ */
+ public DetokenEvaluationErrorListener(OutputStream outputStream) {
+ super(outputStream);
+ }
+
+ @Override
+ public void missclassified(TokenSample reference, TokenSample prediction) {
+ printError(reference, prediction);
+ }
+
+}
diff --git
a/opennlp-tools/src/main/java/opennlp/tools/tokenize/DetokenizerEvaluator.java
b/opennlp-tools/src/main/java/opennlp/tools/tokenize/DetokenizerEvaluator.java
new file mode 100644
index 000000000..7d9df4fc6
--- /dev/null
+++
b/opennlp-tools/src/main/java/opennlp/tools/tokenize/DetokenizerEvaluator.java
@@ -0,0 +1,79 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package opennlp.tools.tokenize;
+
+
+import java.util.ArrayList;
+
+import opennlp.tools.cmdline.tokenizer.DetokenEvaluationErrorListener;
+import opennlp.tools.util.Span;
+import opennlp.tools.util.eval.Evaluator;
+import opennlp.tools.util.eval.FMeasure;
+
+/**
+ * The {@link DetokenizerEvaluator} measures the performance of
+ * the given {@link Detokenizer} with the provided reference
+ * {@link TokenSample}s.
+ *
+ * @see DetokenizerEvaluator
+ * @see Detokenizer
+ * @see TokenSample
+ */
+
+public class DetokenizerEvaluator extends Evaluator<TokenSample> {
+ private FMeasure fmeasure = new FMeasure();
+
+ /**
+ * The {@link Detokenizer} used to create the
+ * predicted tokens.
+ */
+ private Detokenizer detokenizer;
+
+ /**
+ * Initializes the current instance with the
+ * given {@link Detokenizer}.
+ *
+ * @param detokenizer the {@link Detokenizer} to evaluate.
+ * @param listeners evaluation sample listeners
+ */
+ public DetokenizerEvaluator(Detokenizer detokenizer,
DetokenEvaluationErrorListener... listeners) {
+ super(listeners);
+ this.detokenizer = detokenizer;
+ }
+
+ @Override
+ protected TokenSample processSample(TokenSample reference) {
+ String[] tokens = Span.spansToStrings(reference.getTokenSpans(),
reference.getText());
+ String tokensstring = detokenizer.detokenize(tokens, null);
+
+ ArrayList<String> predictionsArray = new ArrayList<>();
+ ArrayList<String> referencesArray = new ArrayList<>();
+
+ predictionsArray.add(tokensstring);
+ referencesArray.add(reference.getText());
+
+ Object[] references = referencesArray.toArray();
+ Object[] predictions = predictionsArray.toArray();
+ fmeasure.updateScores(references, predictions);
+
+ return new TokenSample(tokensstring, reference.getTokenSpans());
+ }
+
+ public FMeasure getFMeasure() {
+ return fmeasure;
+ }
+}
diff --git
a/opennlp-tools/src/test/java/opennlp/tools/tokenize/DetokenizerEvaluatorTest.java
b/opennlp-tools/src/test/java/opennlp/tools/tokenize/DetokenizerEvaluatorTest.java
new file mode 100644
index 000000000..8388cb922
--- /dev/null
+++
b/opennlp-tools/src/test/java/opennlp/tools/tokenize/DetokenizerEvaluatorTest.java
@@ -0,0 +1,81 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package opennlp.tools.tokenize;
+
+import java.io.ByteArrayOutputStream;
+import java.io.OutputStream;
+
+import org.junit.Assert;
+import org.junit.Test;
+
+import opennlp.tools.cmdline.tokenizer.DetokenEvaluationErrorListener;
+import opennlp.tools.util.InvalidFormatException;
+
+
+public class DetokenizerEvaluatorTest {
+ @Test
+ public void testPositive() throws InvalidFormatException {
+ OutputStream stream = new ByteArrayOutputStream();
+ DetokenEvaluationErrorListener listener = new
DetokenEvaluationErrorListener(stream);
+
+ DetokenizerEvaluator eval = new DetokenizerEvaluator(new DummyDetokenizer(
+ TokenSampleTest.createGoldSample()), listener);
+
+ eval.evaluateSample(TokenSampleTest.createGoldSample());
+
+ Assert.assertEquals(1.0, eval.getFMeasure().getFMeasure(), 0.0);
+
+ Assert.assertEquals(0, stream.toString().length());
+ }
+
+ @Test
+ public void testNegative() throws InvalidFormatException {
+ OutputStream stream = new ByteArrayOutputStream();
+ DetokenEvaluationErrorListener listener = new
DetokenEvaluationErrorListener(
+ stream);
+
+ DetokenizerEvaluator eval = new DetokenizerEvaluator(new DummyDetokenizer(
+ TokenSampleTest.createGoldSample()), listener);
+
+ eval.evaluateSample(TokenSampleTest.createPredSilverSample());
+
+ Assert.assertEquals(-1.0d, eval.getFMeasure().getFMeasure(), .1d);
+
+ Assert.assertNotSame(0, stream.toString().length());
+ }
+
+ /**
+ * a dummy tokenizer that always return something expected
+ */
+ class DummyDetokenizer implements Detokenizer {
+
+ private TokenSample sample;
+
+ public DummyDetokenizer(TokenSample sample) {
+ this.sample = sample;
+ }
+
+ public DetokenizationOperation[] detokenize(String[] tokens) {
+ return null;
+ }
+
+ public String detokenize(String[] tokens, String splitMarker) {
+ return this.sample.getText();
+ }
+ }
+}
diff --git
a/opennlp-tools/src/test/java/opennlp/tools/tokenize/TokenSampleTest.java
b/opennlp-tools/src/test/java/opennlp/tools/tokenize/TokenSampleTest.java
index 9b53c8b23..b1dbba289 100644
--- a/opennlp-tools/src/test/java/opennlp/tools/tokenize/TokenSampleTest.java
+++ b/opennlp-tools/src/test/java/opennlp/tools/tokenize/TokenSampleTest.java
@@ -32,12 +32,27 @@
public class TokenSampleTest {
+ public static TokenSample createGoldSample() {
+ return new TokenSample("A test.", new Span[] {new Span(0, 1),
+ new Span(2, 6)});
+ }
+
+ public static TokenSample createPredSample() {
+ return new TokenSample("A test.", new Span[] {new Span(0, 3),
+ new Span(2, 6)});
+ }
+
+ public static TokenSample createPredSilverSample() {
+ return new TokenSample("A t st.", new Span[] {new Span(0, 1),
+ new Span(2, 6)});
+ }
+
@Test
public void testRetrievingContent() {
String sentence = "A test";
- TokenSample sample = new TokenSample(sentence, new Span[]{new Span(0, 1),
+ TokenSample sample = new TokenSample(sentence, new Span[] {new Span(0, 1),
new Span(2, 6)});
Assert.assertEquals("A test", sample.getText());
@@ -75,7 +90,7 @@ public void testCreationWithDetokenizer() throws IOException {
Detokenizer detokenizer =
DictionaryDetokenizerTest.createLatinDetokenizer();
- String[] tokens = new String[]{
+ String[] tokens = new String[] {
"start",
"(", // move right
")", // move left
@@ -116,14 +131,4 @@ public void testEquals() {
Assert.assertFalse(createPredSample().equals(createGoldSample()));
Assert.assertFalse(createPredSample().equals(new Object()));
}
-
- public static TokenSample createGoldSample() {
- return new TokenSample("A test.", new Span[] { new Span(0, 1),
- new Span(2, 6) });
- }
-
- public static TokenSample createPredSample() {
- return new TokenSample("A test.", new Span[] { new Span(0, 3),
- new Span(2, 6) });
- }
}
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
> Add eval support to detokenizer
> -------------------------------
>
> Key: OPENNLP-941
> URL: https://issues.apache.org/jira/browse/OPENNLP-941
> Project: OpenNLP
> Issue Type: Improvement
> Components: Tokenizer
> Reporter: Joern Kottmann
> Priority: Major
> Labels: help-wanted
>
> It would be nice to be able to evaluate a detokenizer on tokenizer training
> data.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
