This is an automated email from the ASF dual-hosted git repository.
rzo1 pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/opennlp.git
The following commit(s) were added to refs/heads/main by this push:
new cf83a1375 OPENNLP-1742: Use the MaxEnt training default for
NameFinderME (#1171)
cf83a1375 is described below
commit cf83a1375c564691acfd22fb6f64be3d67e5861d
Author: jerry317395616 <[email protected]>
AuthorDate: Thu Jul 16 16:28:39 2026 +0800
OPENNLP-1742: Use the MaxEnt training default for NameFinderME (#1171)
* OPENNLP-1742: Use the MaxEnt training default for NameFinderME
* OPENNLP-1742: Stabilize the CLI output test
---
.../opennlp/tools/cmdline/TokenNameFinderToolTest.java | 3 +++
.../main/java/opennlp/tools/namefind/NameFinderME.java | 4 +---
.../java/opennlp/tools/namefind/NameFinderMETest.java | 3 +++
.../tools/namefind/NameFinderMEWithDatesTest.java | 18 ++++++++++++++++--
4 files changed, 23 insertions(+), 5 deletions(-)
diff --git
a/opennlp-core/opennlp-cli/src/test/java/opennlp/tools/cmdline/TokenNameFinderToolTest.java
b/opennlp-core/opennlp-cli/src/test/java/opennlp/tools/cmdline/TokenNameFinderToolTest.java
index da3d9e1fa..4806cf8fa 100644
---
a/opennlp-core/opennlp-cli/src/test/java/opennlp/tools/cmdline/TokenNameFinderToolTest.java
+++
b/opennlp-core/opennlp-cli/src/test/java/opennlp/tools/cmdline/TokenNameFinderToolTest.java
@@ -37,6 +37,7 @@ import org.junit.jupiter.api.Test;
import org.slf4j.LoggerFactory;
import opennlp.tools.cmdline.namefind.TokenNameFinderTool;
+import opennlp.tools.ml.AlgorithmType;
import opennlp.tools.namefind.NameFinderME;
import opennlp.tools.namefind.NameSample;
import opennlp.tools.namefind.NameSampleDataStream;
@@ -128,6 +129,8 @@ public class TokenNameFinderToolTest {
StandardCharsets.ISO_8859_1);
TrainingParameters params = new TrainingParameters();
+ // This test verifies CLI output, so keep its fixture independent of the
default trainer.
+ params.put(Parameters.ALGORITHM_PARAM,
AlgorithmType.PERCEPTRON.getAlgorithmType());
params.put(Parameters.ITERATIONS_PARAM, 70);
params.put(Parameters.CUTOFF_PARAM, 1);
diff --git
a/opennlp-core/opennlp-runtime/src/main/java/opennlp/tools/namefind/NameFinderME.java
b/opennlp-core/opennlp-runtime/src/main/java/opennlp/tools/namefind/NameFinderME.java
index f02feace5..f1dec349c 100644
---
a/opennlp-core/opennlp-runtime/src/main/java/opennlp/tools/namefind/NameFinderME.java
+++
b/opennlp-core/opennlp-runtime/src/main/java/opennlp/tools/namefind/NameFinderME.java
@@ -28,7 +28,6 @@ import java.util.regex.Matcher;
import java.util.regex.Pattern;
import opennlp.tools.commons.ThreadSafe;
-import opennlp.tools.ml.AlgorithmType;
import opennlp.tools.ml.BeamSearch;
import opennlp.tools.ml.EventModelSequenceTrainer;
import opennlp.tools.ml.EventTrainer;
@@ -273,8 +272,7 @@ public class NameFinderME implements TokenNameFinder,
Probabilistic {
ObjectStream<NameSample> samples,
TrainingParameters params,
TokenNameFinderFactory factory)
throws IOException {
- //FIXME OPENNLP-1742
- params.putIfAbsent(Parameters.ALGORITHM_PARAM,
AlgorithmType.PERCEPTRON.getAlgorithmType());
+ params.putIfAbsent(Parameters.ALGORITHM_PARAM,
Parameters.ALGORITHM_DEFAULT_VALUE);
params.putIfAbsent(Parameters.CUTOFF_PARAM, 0);
params.putIfAbsent(Parameters.ITERATIONS_PARAM, 300);
diff --git
a/opennlp-core/opennlp-runtime/src/test/java/opennlp/tools/namefind/NameFinderMETest.java
b/opennlp-core/opennlp-runtime/src/test/java/opennlp/tools/namefind/NameFinderMETest.java
index b014e06c6..3c8c8e6bc 100644
---
a/opennlp-core/opennlp-runtime/src/test/java/opennlp/tools/namefind/NameFinderMETest.java
+++
b/opennlp-core/opennlp-runtime/src/test/java/opennlp/tools/namefind/NameFinderMETest.java
@@ -76,6 +76,9 @@ public class NameFinderMETest extends AbstractNameFinderTest {
TokenNameFinderModel nameFinderModel = NameFinderME.train("eng", null,
sampleStream,
params, TokenNameFinderFactory.create(null, null,
Collections.emptyMap(), new BioCodec()));
+ assertEquals(Parameters.ALGORITHM_DEFAULT_VALUE,
+ params.getStringParameter(Parameters.ALGORITHM_PARAM, null));
+
TokenNameFinder nameFinder = new NameFinderME(nameFinderModel);
// now test if it can detect the sample sentences
diff --git
a/opennlp-core/opennlp-runtime/src/test/java/opennlp/tools/namefind/NameFinderMEWithDatesTest.java
b/opennlp-core/opennlp-runtime/src/test/java/opennlp/tools/namefind/NameFinderMEWithDatesTest.java
index e6abb6a53..9e3cb1720 100644
---
a/opennlp-core/opennlp-runtime/src/test/java/opennlp/tools/namefind/NameFinderMEWithDatesTest.java
+++
b/opennlp-core/opennlp-runtime/src/test/java/opennlp/tools/namefind/NameFinderMEWithDatesTest.java
@@ -28,7 +28,10 @@ import org.junit.jupiter.params.provider.Arguments;
import org.junit.jupiter.params.provider.MethodSource;
import org.junit.jupiter.params.provider.ValueSource;
+import opennlp.tools.ml.AlgorithmType;
+import opennlp.tools.util.Parameters;
import opennlp.tools.util.Span;
+import opennlp.tools.util.TrainingParameters;
import static org.junit.jupiter.api.Assertions.assertEquals;
import static org.junit.jupiter.api.Assertions.assertNotNull;
@@ -69,11 +72,11 @@ public class NameFinderMEWithDatesTest extends
AbstractNameFinderTest {
downloadVersion15Model("en-ner-date.bin");
v15ModelEN = new TokenNameFinderModel(Files.newInputStream(
OPENNLP_DIR.resolve("en-ner-date.bin")));
- TokenNameFinderModel trainedModelEN = trainModel("eng",
+ TokenNameFinderModel trainedModelEN = trainDateModel("eng",
"opennlp/tools/namefind/RandomNewsWithGeneratedDates_EN.train");
assertNotNull(trainedModelEN);
assertTrue(hasOtherAsOutcome(trainedModelEN));
- TokenNameFinderModel trainedModelDE = trainModel("deu",
+ TokenNameFinderModel trainedModelDE = trainDateModel("deu",
"opennlp/tools/namefind/RandomNewsWithGeneratedDates_DE.train");
assertNotNull(trainedModelDE);
assertTrue(hasOtherAsOutcome(trainedModelDE));
@@ -84,6 +87,17 @@ public class NameFinderMEWithDatesTest extends
AbstractNameFinderTest {
}
}
+ private static TokenNameFinderModel trainDateModel(String languageCode,
+ String trainingFile)
throws IOException {
+ TrainingParameters params = new TrainingParameters();
+ // This suite verifies date spans, so keep its established trainer
independent of the default.
+ params.put(Parameters.ALGORITHM_PARAM,
AlgorithmType.PERCEPTRON.getAlgorithmType());
+ params.put(Parameters.ITERATIONS_PARAM, 150);
+ params.put(Parameters.THREADS_PARAM, 4);
+ params.put(Parameters.CUTOFF_PARAM, 3);
+ return trainModel(languageCode, trainingFile, params);
+ }
+
/**
* Uses the English SourceForge V1.5 model with "date" nameType and verifies
with sample text.
*/
