Merge pull request #4394 from inception-project/feature/2696-Document…

…-level-recommendations #2696 - Document-level recommendations
inception-project · Dec 27, 2023 · 30ea848 · 30ea848
2 parents 5ec948d + 9e04269
commit 30ea848
Show file tree

Hide file tree

Showing 19 changed files with 300 additions and 348 deletions.
diff --git a/...r/src/test/java/de/tudarmstadt/ukp/clarin/webanno/brat/render/BratSerializerImplTest.java b/...r/src/test/java/de/tudarmstadt/ukp/clarin/webanno/brat/render/BratSerializerImplTest.java
@@ -23,6 +23,8 @@
 import static java.util.Arrays.asList;
 import static org.apache.uima.fit.factory.AnalysisEngineFactory.createEngine;
 import static org.apache.uima.fit.factory.CollectionReaderFactory.createReader;
+import static org.apache.uima.fit.util.CasUtil.getType;
+import static org.apache.uima.fit.util.CasUtil.select;
 import static org.assertj.core.api.Assertions.assertThat;
 import static org.assertj.core.api.Assertions.contentOf;
 import static org.mockito.ArgumentMatchers.any;
@@ -32,6 +34,7 @@
 
 import org.apache.uima.analysis_engine.AnalysisEngine;
 import org.apache.uima.cas.CAS;
+import org.apache.uima.cas.text.AnnotationFS;
 import org.apache.uima.collection.CollectionReader;
 import org.apache.uima.fit.factory.JCasFactory;
 import org.dkpro.core.io.tcf.TcfReader;
@@ -59,6 +62,7 @@
 import de.tudarmstadt.ukp.clarin.webanno.model.Project;
 import de.tudarmstadt.ukp.clarin.webanno.model.SourceDocument;
 import de.tudarmstadt.ukp.dkpro.core.api.lexmorph.type.pos.POS;
+import de.tudarmstadt.ukp.dkpro.core.api.segmentation.type.Sentence;
 import de.tudarmstadt.ukp.dkpro.core.api.segmentation.type.Token;
 import de.tudarmstadt.ukp.inception.annotation.feature.bool.BooleanFeatureSupport;
 import de.tudarmstadt.ukp.inception.annotation.feature.link.LinkFeatureSupport;
@@ -76,7 +80,6 @@
 import de.tudarmstadt.ukp.inception.schema.api.AnnotationSchemaService;
 import de.tudarmstadt.ukp.inception.schema.service.FeatureSupportRegistryImpl;
 import de.tudarmstadt.ukp.inception.support.json.JSONUtil;
-import de.tudarmstadt.ukp.inception.support.uima.WebAnnoCasUtil;
 
 @ExtendWith(MockitoExtension.class)
 public class BratSerializerImplTest
@@ -180,7 +183,7 @@ public void thatSentenceOrientedStrategyRenderCorrectly() throws Exception
         state.setAllAnnotationLayers(schemaService.listAnnotationLayer(project));
         state.setPagingStrategy(new SentenceOrientedPagingStrategy());
         state.getPreferences().setWindowSize(10);
-        state.setFirstVisibleUnit(WebAnnoCasUtil.getFirstSentence(cas));
+        state.setFirstVisibleUnit(getFirstSentence(cas));
         state.setProject(project);
         state.setDocument(sourceDocument, asList(sourceDocument));
 
@@ -219,7 +222,7 @@ public void thatLineOrientedStrategyRenderCorrectly() throws Exception
         AnnotatorState state = new AnnotatorStateImpl(Mode.ANNOTATION);
         state.setPagingStrategy(new LineOrientedPagingStrategy());
         state.getPreferences().setWindowSize(10);
-        state.setFirstVisibleUnit(WebAnnoCasUtil.getFirstSentence(cas));
+        state.setFirstVisibleUnit(getFirstSentence(cas));
         state.setProject(project);
         state.setDocument(sourceDocument, asList(sourceDocument));
 
@@ -258,7 +261,7 @@ public void thatTokenWrappingStrategyRenderCorrectly() throws Exception
         AnnotatorState state = new AnnotatorStateImpl(Mode.ANNOTATION);
         state.setPagingStrategy(new TokenWrappingPagingStrategy(80));
         state.getPreferences().setWindowSize(10);
-        state.setFirstVisibleUnit(WebAnnoCasUtil.getFirstSentence(cas));
+        state.setFirstVisibleUnit(getFirstSentence(cas));
         state.setProject(project);
         state.setDocument(sourceDocument, asList(sourceDocument));
 
@@ -281,4 +284,22 @@ public void thatTokenWrappingStrategyRenderCorrectly() throws Exception
         assertThat(contentOf(new File("src/test/resources/longlines.json"), UTF_8))
                 .isEqualToNormalizingNewlines(contentOf(new File(jsonFilePath), UTF_8));
     }
+
+    /**
+     * Get the internal address of the first sentence annotation from CAS. This will be used as a
+     * reference for moving forward/backward sentences positions
+     *
+     * @param aCas
+     *            The CAS object assumed to contains some sentence annotations
+     * @return the sentence number or -1 if aCas don't have sentence annotation
+     */
+    private static AnnotationFS getFirstSentence(CAS aCas)
+    {
+        AnnotationFS firstSentence = null;
+        for (AnnotationFS s : select(aCas, getType(aCas, Sentence.class))) {
+            firstSentence = s;
+            break;
+        }
+        return firstSentence;
+    }
 }
diff --git a/...n/java/de/tudarmstadt/ukp/inception/conceptlinking/service/ConceptLinkingServiceImpl.java b/...n/java/de/tudarmstadt/ukp/inception/conceptlinking/service/ConceptLinkingServiceImpl.java
@@ -24,13 +24,13 @@
 import static de.tudarmstadt.ukp.inception.conceptlinking.model.CandidateEntity.KEY_QUERY;
 import static de.tudarmstadt.ukp.inception.conceptlinking.model.CandidateEntity.KEY_QUERY_BEST_MATCH_TERM_NC;
 import static de.tudarmstadt.ukp.inception.conceptlinking.model.CandidateEntity.KEY_QUERY_NC;
-import static de.tudarmstadt.ukp.inception.support.uima.WebAnnoCasUtil.selectSentenceCovering;
-import static de.tudarmstadt.ukp.inception.support.uima.WebAnnoCasUtil.selectTokensCovered;
 import static java.lang.System.currentTimeMillis;
 import static java.util.Arrays.asList;
 import static java.util.Collections.unmodifiableList;
 import static java.util.Comparator.comparingInt;
 import static java.util.stream.Collectors.toCollection;
+import static org.apache.uima.fit.util.CasUtil.getType;
+import static org.apache.uima.fit.util.CasUtil.select;
 
 import java.io.File;
 import java.net.URISyntaxException;
@@ -47,6 +47,7 @@
 import org.apache.commons.lang3.Validate;
 import org.apache.uima.cas.CAS;
 import org.apache.uima.cas.text.AnnotationFS;
+import org.apache.uima.fit.util.CasUtil;
 import org.eclipse.rdf4j.common.net.ParsedIRI;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
@@ -58,6 +59,8 @@
 import org.springframework.core.annotation.AnnotationAwareOrderComparator;
 
 import de.tudarmstadt.ukp.clarin.webanno.model.Project;
+import de.tudarmstadt.ukp.dkpro.core.api.segmentation.type.Sentence;
+import de.tudarmstadt.ukp.dkpro.core.api.segmentation.type.Token;
 import de.tudarmstadt.ukp.inception.conceptlinking.config.EntityLinkingProperties;
 import de.tudarmstadt.ukp.inception.conceptlinking.config.EntityLinkingPropertiesImpl;
 import de.tudarmstadt.ukp.inception.conceptlinking.config.EntityLinkingServiceAutoConfiguration;
@@ -513,4 +516,31 @@ public List<KBHandle> searchItems(KnowledgeBase aKB, String aQuery)
     {
         return disambiguate(aKB, null, ConceptFeatureValueType.ANY_OBJECT, aQuery, null, 0, null);
     }
+
+    /**
+     * Get the sentence based on the annotation begin offset
+     *
+     * @param aCas
+     *            the CAS.
+     * @param aBegin
+     *            the begin offset.
+     * @return the sentence.
+     */
+    private static AnnotationFS selectSentenceCovering(CAS aCas, int aBegin)
+    {
+        AnnotationFS currentSentence = null;
+        for (AnnotationFS sentence : select(aCas, getType(aCas, Sentence.class))) {
+            if (sentence.getBegin() <= aBegin && sentence.getEnd() > aBegin) {
+                currentSentence = sentence;
+                break;
+            }
+        }
+        return currentSentence;
+    }
+
+    private static Collection<AnnotationFS> selectTokensCovered(AnnotationFS aCover)
+    {
+        return CasUtil.selectCovered(aCover.getCAS(), getType(aCover.getCAS(), Token.class),
+                aCover);
+    }
 }
diff --git a/.../main/java/de/tudarmstadt/ukp/clarin/webanno/diag/checks/UnreachableAnnotationsCheck.java b/.../main/java/de/tudarmstadt/ukp/clarin/webanno/diag/checks/UnreachableAnnotationsCheck.java
@@ -24,10 +24,13 @@
 import static org.apache.uima.cas.impl.Serialization.deserializeCASComplete;
 import static org.apache.uima.cas.impl.Serialization.serializeCASComplete;
 
+import java.util.LinkedHashSet;
 import java.util.List;
 import java.util.Map;
+import java.util.Set;
 
 import org.apache.uima.cas.CAS;
+import org.apache.uima.cas.FeatureStructure;
 import org.apache.uima.cas.impl.CASImpl;
 import org.apache.uima.resource.ResourceInitializationException;
 
@@ -93,8 +96,15 @@ public static CASImpl makeDummyCas()
 
     public static Map<String, Long> countFeatureStructures(CASImpl casImpl)
     {
-        return WebAnnoCasUtil.findAllFeatureStructures(casImpl).stream() //
+        return findAllFeatureStructures(casImpl).stream() //
                 .map(fs -> fs.getType().getName()) //
                 .collect(groupingBy(identity(), counting()));
     }
+
+    public static Set<FeatureStructure> findAllFeatureStructures(CAS aCas)
+    {
+        Set<FeatureStructure> allFSes = new LinkedHashSet<>();
+        ((CASImpl) aCas).walkReachablePlusFSsSorted(allFSes::add, null, null, null);
+        return allFSes;
+    }
 }
diff --git a/...a/de/tudarmstadt/ukp/inception/recommendation/imls/opennlp/ner/OpenNlpNerRecommender.java b/...a/de/tudarmstadt/ukp/inception/recommendation/imls/opennlp/ner/OpenNlpNerRecommender.java
@@ -104,10 +104,10 @@ public void train(RecommenderContext aContext, List<CAS> aCasses) throws Recomme
         // OpenNLP
         int beamSize = Math.max(maxRecommendations, NameFinderME.DEFAULT_BEAM_SIZE);
 
-        TrainingParameters params = traits.getParameters();
+        var params = traits.getParameters();
         params.put(BeamSearch.BEAM_SIZE_PARAMETER, Integer.toString(beamSize));
 
-        TokenNameFinderModel model = train(nameSamples, params);
+        var model = train(nameSamples, params);
 
         aContext.put(KEY_MODEL, model);
     }
@@ -304,7 +304,7 @@ private String determineLabel(Span aName, int aTokenIdx)
         return label;
     }
 
-    private List<NameSample> extractNameSamples(List<CAS> aCasses)
+    private List<NameSample> extractNameSamples(Iterable<CAS> aCasses)
     {
         var nameSamples = new ArrayList<NameSample>();
 
@@ -402,8 +402,8 @@ private TokenNameFinderModel train(List<NameSample> aNameSamples,
             TrainingParameters aParameters)
         throws RecommendationException
     {
-        try (NameSampleStream stream = new NameSampleStream(aNameSamples)) {
-            TokenNameFinderFactory finderFactory = new TokenNameFinderFactory();
+        try (var stream = new NameSampleStream(aNameSamples)) {
+            var finderFactory = new TokenNameFinderFactory();
             return NameFinderME.train("unknown", null, stream, aParameters, finderFactory);
         }
         catch (IOException e) {

diff --git a/...e/tudarmstadt/ukp/inception/ui/core/docanno/recommendation/MetadataSuggestionSupport.java b/...e/tudarmstadt/ukp/inception/ui/core/docanno/recommendation/MetadataSuggestionSupport.java
@@ -19,8 +19,6 @@
 
 import static de.tudarmstadt.ukp.inception.recommendation.api.model.AnnotationSuggestion.FLAG_ALL;
 import static de.tudarmstadt.ukp.inception.recommendation.api.model.AnnotationSuggestion.FLAG_OVERLAP;
-import static de.tudarmstadt.ukp.inception.recommendation.api.model.AnnotationSuggestion.FLAG_TRANSIENT_REJECTED;
-import static de.tudarmstadt.ukp.inception.recommendation.api.model.LearningRecordUserAction.REJECTED;
 
 import java.lang.invoke.MethodHandles;
 import java.util.ArrayList;
@@ -29,7 +27,6 @@
 import java.util.Objects;
 import java.util.Optional;
 
-import org.apache.commons.lang3.NotImplementedException;
 import org.apache.uima.cas.AnnotationBaseFS;
 import org.apache.uima.cas.CAS;
 import org.apache.uima.cas.Feature;
@@ -45,7 +42,6 @@
 import de.tudarmstadt.ukp.inception.recommendation.api.RecommendationService;
 import de.tudarmstadt.ukp.inception.recommendation.api.SuggestionRenderer;
 import de.tudarmstadt.ukp.inception.recommendation.api.SuggestionSupport_ImplBase;
-import de.tudarmstadt.ukp.inception.recommendation.api.event.RecommendationRejectedEvent;
 import de.tudarmstadt.ukp.inception.recommendation.api.model.AnnotationSuggestion;
 import de.tudarmstadt.ukp.inception.recommendation.api.model.LearningRecord;
 import de.tudarmstadt.ukp.inception.recommendation.api.model.LearningRecordChangeLocation;
@@ -135,52 +131,22 @@ else if (candidates.isEmpty() || !aAdapter.getTraits(DocumentMetadataLayerTraits
         return annotation;
     }
 
-    @Override
-    public void rejectSuggestion(String aSessionOwner, SourceDocument aDocument, String aDataOwner,
-            AnnotationSuggestion aSuggestion, LearningRecordChangeLocation aAction)
-        throws AnnotationException
-    {
-        var suggestion = (MetadataSuggestion) aSuggestion;
-
-        // Hide the suggestion. This is faster than having to recalculate the visibility status
-        // for the entire document or even for the part visible on screen.
-        suggestion.hide(FLAG_TRANSIENT_REJECTED);
-
-        var recommender = recommendationService.getRecommender(suggestion);
-        var feature = recommender.getFeature();
-        // Log the action to the learning record
-        learningRecordService.logRecord(aSessionOwner, aDocument, aDataOwner, suggestion, feature,
-                REJECTED, aAction);
-
-        // Send an application event that the suggestion has been rejected
-        applicationEventPublisher.publishEvent(new RecommendationRejectedEvent(this, aDocument,
-                aDataOwner, feature, suggestion.getLabel()));
-    }
-
-    @Override
-    public void skipSuggestion(String aSessionOwner, SourceDocument aDocument, String aDataOwner,
-            AnnotationSuggestion aSuggestion, LearningRecordChangeLocation aAction)
-        throws AnnotationException
-    {
-        throw new NotImplementedException("Not yet implemented");
-    }
-
     @Override
     public <T extends AnnotationSuggestion> void calculateSuggestionVisibility(String aSessionOwner,
             SourceDocument aDocument, CAS aCas, String aDataOwner, AnnotationLayer aLayer,
             Collection<SuggestionGroup<T>> aRecommendations, int aWindowBegin, int aWindowEnd)
     {
         LOG.trace("calculateSuggestionVisibility() for layer {} on document {}", aLayer, aDocument);
 
-        var type = aCas.getTypeSystem().getType(aLayer.getName());
-        if (type == null) {
+        var predictedType = aCas.getTypeSystem().getType(aLayer.getName());
+        if (predictedType == null) {
             // The type does not exist in the type system of the CAS. Probably it has not
             // been upgraded to the latest version of the type system yet. If this is the case,
             // we'll just skip.
             return;
         }
 
-        var annotations = aCas.<AnnotationBase> select(type).asList();
+        var annotations = aCas.<AnnotationBase> select(predictedType).asList();
 
         var suggestionsForLayer = aRecommendations.stream()
                 // Only suggestions for the given layer
@@ -194,7 +160,7 @@ public <T extends AnnotationSuggestion> void calculateSuggestionVisibility(Strin
         var adapter = schemaService.getAdapter(aLayer);
         var traits = adapter.getTraits(DocumentMetadataLayerTraits.class).get();
         for (var feature : schemaService.listSupportedFeatures(aLayer)) {
-            var feat = type.getFeatureByBaseName(feature.getName());
+            var feat = predictedType.getFeatureByBaseName(feature.getName());
 
             if (feat == null) {
                 // The feature does not exist in the type system of the CAS. Probably it has not
@@ -267,14 +233,16 @@ static void hideSuggestionsRejectedOrSkipped(MetadataSuggestion aSuggestion,
     }
 
     @Override
-    public LearningRecord toLearningRecord(SourceDocument aDocument, String aUsername,
+    public LearningRecord toLearningRecord(SourceDocument aDocument, String aDataOwner,
             AnnotationSuggestion aSuggestion, AnnotationFeature aFeature,
             LearningRecordUserAction aUserAction, LearningRecordChangeLocation aLocation)
     {
         var record = new LearningRecord();
-        record.setUser(aUsername);
+        record.setUser(aDataOwner);
         record.setSourceDocument(aDocument);
         record.setUserAction(aUserAction);
+        record.setOffsetBegin(-1);
+        record.setOffsetEnd(-1);
         record.setOffsetBegin2(-1);
         record.setOffsetEnd2(-1);
         record.setAnnotation(aSuggestion.getLabel());

diff --git a/inception/inception-pdf-editor/pom.xml b/inception/inception-pdf-editor/pom.xml
@@ -87,6 +87,10 @@
       <groupId>org.dkpro.core</groupId>
       <artifactId>dkpro-core-io-pdf-asl</artifactId>
     </dependency>
+    <dependency>
+      <groupId>org.dkpro.core</groupId>
+      <artifactId>dkpro-core-api-segmentation-asl</artifactId>
+    </dependency>
     <dependency>
       <groupId>org.dkpro.core</groupId>
       <artifactId>dkpro-core-api-resources-asl</artifactId>
@@ -168,11 +172,6 @@
       <artifactId>dkpro-core-api-lexmorph-asl</artifactId>
       <scope>test</scope>
     </dependency>
-    <dependency>
-      <groupId>org.dkpro.core</groupId>
-      <artifactId>dkpro-core-api-segmentation-asl</artifactId>
-      <scope>test</scope>
-    </dependency>
     <dependency>
       <groupId>de.tudarmstadt.ukp.inception.app</groupId>
       <artifactId>inception-schema</artifactId>

diff --git a/...-pdf-editor/src/main/java/de/tudarmstadt/ukp/inception/pdfeditor/PdfAnnotationEditor.java b/...-pdf-editor/src/main/java/de/tudarmstadt/ukp/inception/pdfeditor/PdfAnnotationEditor.java
@@ -20,10 +20,10 @@
 import static de.tudarmstadt.ukp.inception.pdfeditor.pdfanno.render.PdfAnnoSerializer.convertToDocumentOffset;
 import static de.tudarmstadt.ukp.inception.pdfeditor.pdfanno.render.PdfAnnoSerializer.convertToDocumentOffsets;
 import static de.tudarmstadt.ukp.inception.rendering.vmodel.VID.NONE_ID;
-import static de.tudarmstadt.ukp.inception.support.uima.WebAnnoCasUtil.selectSentenceAt;
 import static de.tudarmstadt.ukp.inception.support.wicket.ServletContextUtils.referenceToUrl;
 import static java.lang.String.join;
 import static java.util.Arrays.asList;
+import static org.apache.uima.fit.util.CasUtil.getType;
 
 import java.io.IOException;
 import java.util.ArrayList;
@@ -37,6 +37,7 @@
 import org.apache.commons.lang3.exception.ExceptionUtils;
 import org.apache.uima.cas.CAS;
 import org.apache.uima.cas.text.AnnotationFS;
+import org.apache.uima.fit.util.CasUtil;
 import org.apache.wicket.Component;
 import org.apache.wicket.ajax.AjaxRequestTarget;
 import org.apache.wicket.feedback.IFeedback;
@@ -48,6 +49,7 @@
 import org.slf4j.LoggerFactory;
 
 import de.tudarmstadt.ukp.clarin.webanno.api.casstorage.CasProvider;
+import de.tudarmstadt.ukp.dkpro.core.api.segmentation.type.Sentence;
 import de.tudarmstadt.ukp.inception.documents.api.DocumentService;
 import de.tudarmstadt.ukp.inception.editor.AnnotationEditorExtensionRegistry;
 import de.tudarmstadt.ukp.inception.editor.action.AnnotationActionHandler;
@@ -314,4 +316,20 @@ private String getAnnotationsJS(PdfAnnoModel aPdfAnnoModel)
                 "'colorMap': {},", //
                 "'annotations':[annoFile]}, true);");
     }
+
+    /**
+     * Get the sentence for this CAS based on the begin and end offsets. This is basically used to
+     * transform sentence address in one CAS to other sentence address for different CAS
+     *
+     * @param aCas
+     *            the CAS.
+     * @param aBegin
+     *            the begin offset.
+     * @return the sentence.
+     */
+    private static AnnotationFS selectSentenceAt(CAS aCas, int aBegin)
+    {
+        return CasUtil.select(aCas, getType(aCas, Sentence.class)).stream()
+                .filter(s -> s.getBegin() == aBegin).findFirst().orElse(null);
+    }
 }