#250, #344 - added option to upload dataset and answer file

dice-group · Jul 14, 2020 · d526c11 · d526c11
1 parent 0b4b28c
commit d526c11
Show file tree

Hide file tree

Showing 10 changed files with 179 additions and 110 deletions.
diff --git a/src/main/java/org/aksw/gerbil/annotator/InstanceListBasedConfigurationImpl.java b/src/main/java/org/aksw/gerbil/annotator/InstanceListBasedConfigurationImpl.java
@@ -65,6 +65,7 @@ public Annotator getAnnotator(ExperimentType experimentType) throws GerbilExcept
     }
 
     protected Annotator loadAnnotator(ExperimentType experimentType) throws Exception {
+        datasetConfig.setQuestionLanguage(this.questionLang);
         Dataset dataset = datasetConfig.getDataset(experimentType);
         if (dataset == null) {
             return null;

diff --git a/src/main/java/org/aksw/gerbil/dataset/AbstractDatasetConfiguration.java b/src/main/java/org/aksw/gerbil/dataset/AbstractDatasetConfiguration.java
@@ -33,19 +33,38 @@ public abstract class AbstractDatasetConfiguration extends AbstractAdapterConfig
     protected EntityCheckerManager entityCheckerManager;
     protected SameAsRetriever globalRetriever;
     protected String questionLang;
+    protected String questionLabel;
 
-    public AbstractDatasetConfiguration(String datasetName, boolean couldBeCached,
+    public AbstractDatasetConfiguration(String datasetName,boolean couldBeCached,
             ExperimentType applicableForExperiment, EntityCheckerManager entityCheckerManager,
             SameAsRetriever globalRetriever) {
         super(datasetName, couldBeCached, applicableForExperiment);
+        this.questionLabel=datasetName;
         this.entityCheckerManager = entityCheckerManager;
         this.globalRetriever = globalRetriever;
     }
-
+
+    public AbstractDatasetConfiguration(String datasetName, String questionLabel, boolean couldBeCached,
+                                        ExperimentType applicableForExperiment, EntityCheckerManager entityCheckerManager,
+                                        SameAsRetriever globalRetriever) {
+        super(datasetName, couldBeCached, applicableForExperiment);
+        this.questionLabel=questionLabel;
+        this.entityCheckerManager = entityCheckerManager;
+        this.globalRetriever = globalRetriever;
+    }
+
     public void setQuestionLang(String questionLang){
     	this.questionLang = questionLang;
     }
 
+    public String getQuestionLabel() {
+        return questionLabel;
+    }
+
+    public void setQuestionLabel(String questionLabel) {
+        this.questionLabel = questionLabel;
+    }
+
     @Override
     public Dataset getDataset(ExperimentType experimentType) throws GerbilException {
         // for (int i = 0; i < applicableForExperiments.length; ++i) {

diff --git a/src/main/java/org/aksw/gerbil/dataset/Dataset.java b/src/main/java/org/aksw/gerbil/dataset/Dataset.java
@@ -37,4 +37,8 @@ public interface Dataset extends Closeable {
     public void setQuestionLanguage(String qLang);
 
     public String getQuestionLanguage();
+
+    public String getQuestionLabel();
+
+    public void setQuestionLabel(String questionLabel);
 }
diff --git a/src/main/java/org/aksw/gerbil/dataset/impl/AbstractDataset.java b/src/main/java/org/aksw/gerbil/dataset/impl/AbstractDataset.java
@@ -24,6 +24,7 @@
 public abstract class AbstractDataset implements Dataset {
 
     protected String name;
+    protected String questionLabel;
     protected ClosePermitionGranter granter;
     protected String qLang;
 
@@ -33,6 +34,23 @@ public AbstractDataset() {
 
     public AbstractDataset(String name) {
         this.name = name;
+        this.questionLabel=name;
+    }
+
+
+    public AbstractDataset(String name, String questionLabel) {
+        this.name = name;
+        this.questionLabel=questionLabel;
+    }
+
+    @Override
+    public String getQuestionLabel() {
+        return questionLabel;
+    }
+
+    @Override
+    public void setQuestionLabel(String questionLabel) {
+        this.questionLabel = questionLabel;
     }
 
     @Override

diff --git a/src/main/java/org/aksw/gerbil/dataset/impl/qald/FileBasedQALDDataset.java b/src/main/java/org/aksw/gerbil/dataset/impl/qald/FileBasedQALDDataset.java
@@ -17,6 +17,7 @@
 import org.aksw.gerbil.qa.QALDStreamUtils;
 import org.aksw.gerbil.qa.QAUtils;
 import org.aksw.gerbil.transfer.nif.Document;
+import org.aksw.gerbil.web.config.AdapterManager;
 import org.aksw.qa.commons.datastructure.IQuestion;
 import org.aksw.qa.commons.load.LoaderController;
 import org.aksw.qa.commons.load.json.EJQuestionFactory;
@@ -40,8 +41,8 @@ public FileBasedQALDDataset(String file){
     	this.file=file;
     	initLanguage();
     }
-    public FileBasedQALDDataset(String name, String file){
-    	super(name);
+    public FileBasedQALDDataset(String name, String questionLabel, String file){
+    	super(name, questionLabel);
     	this.file=file;
     	initLanguage();
     }
@@ -52,14 +53,14 @@ public FileBasedQALDDataset(String file, QALDStreamType fileType) {
         initLanguage();
     }
 
-    public FileBasedQALDDataset(String name, String file, QALDStreamType fileType) {
-        super(name);
+    public FileBasedQALDDataset(String name, String questionLabel, String file, QALDStreamType fileType) {
+        super(name, questionLabel);
         this.file = file;
         this.fileType = fileType;
         initLanguage();
     }
-    public FileBasedQALDDataset(String name, String file, String questionLanguage) {
-        super(name);
+    public FileBasedQALDDataset(String name, String questionLabel, String file, String questionLanguage) {
+        super(name, questionLabel);
         this.file = file;
         this.fileType = fileType;
         this.qLang=questionLanguage;
@@ -86,7 +87,7 @@ public void init() throws GerbilException {
         try {
             fin = new FileInputStream(file);
             if(fileType!=null){
-            	instances = QALDStreamUtils.parseDocument(fin, fileType, getName());
+            	instances = QALDStreamUtils.parseDocument(fin, fileType, getQuestionLabel());
             }
             else{
         		List<IQuestion> questions;
@@ -96,7 +97,7 @@ public void init() throws GerbilException {
         			//XML
         			questions = LoaderController.loadXML(fin, null, qLang);
         		}
-        		instances = generateInstancesFromQuestions(getName(), questions, qLang);
+        		instances = generateInstancesFromQuestions(getQuestionLabel(), questions, qLang);
             }
         } catch (Exception e) {
         	IOUtils.closeQuietly(fin);

diff --git a/src/main/java/org/aksw/gerbil/dataset/impl/qald/QALDFileDatasetConfig.java b/src/main/java/org/aksw/gerbil/dataset/impl/qald/QALDFileDatasetConfig.java
@@ -34,9 +34,16 @@ public QALDFileDatasetConfig(String name, String file, boolean couldBeCached, Ex
         this.file = file;
     }
 
+
+    public QALDFileDatasetConfig(String name, String questionLabel, String file, boolean couldBeCached, ExperimentType applicableForExperiment,
+                                 EntityCheckerManager entityCheckerManager, SameAsRetriever globalRetriever) {
+        super(name, questionLabel, couldBeCached, applicableForExperiment, entityCheckerManager, globalRetriever);
+        this.file = file;
+    }
+
     @Override
     protected Dataset loadDataset() throws Exception {
-        FileBasedQALDDataset dataset = new FileBasedQALDDataset(getName(), file);
+        FileBasedQALDDataset dataset = new FileBasedQALDDataset(getName(), getQuestionLabel(), file, questionLang);
         dataset.setQuestionLanguage(questionLang);
         dataset.init();
         return dataset;

diff --git a/src/main/java/org/aksw/gerbil/execute/ExperimentTask.java b/src/main/java/org/aksw/gerbil/execute/ExperimentTask.java
@@ -248,7 +248,7 @@ protected void prepareAnnotatorResults(List<? extends List<? extends Meaning>> r
 		case Sc2KB:
 		case OKE_Task1: // falls through
 		case OKE_Task2:
-		case QA:
+
 		case ETyping: {
 			if (annotatorSameAsRetriever != null) {
 				for (List<? extends Meaning> result : results) {
@@ -257,6 +257,7 @@ protected void prepareAnnotatorResults(List<? extends List<? extends Meaning>> r
 			}
 			return;
 		}
+		case QA:
 		case ERec:// falls through
 		default:
 			// nothing to do

diff --git a/src/main/java/org/aksw/gerbil/web/MainController.java b/src/main/java/org/aksw/gerbil/web/MainController.java
@@ -190,6 +190,7 @@ public ModelAndView index() {
         for (String answerFile : answerFiles) {
             configs[count] = new ExperimentTaskConfiguration(adapterManager.getAnnotatorConfig(answerFile, type, qLang),
                     adapterManager.getDatasetConfig(answerFile, type, qLang), type, getMatching(matching), qLang);
+
             LOGGER.debug("Created config: {}", configs[count]);
             ++count;
         }

diff --git a/src/main/java/org/aksw/gerbil/web/config/AdapterManager.java b/src/main/java/org/aksw/gerbil/web/config/AdapterManager.java
@@ -48,7 +48,8 @@ public class AdapterManager {
     private static final String NIF_WS_PREFIX = "NIFWS_";
     private static final String NIF_WS_SUFFIX = " (WS)";
     private static final String AF_PREFIX = "AF_";
-    private static final String UPLOADED_AF_SUFFIX = " (uploaded)";
+    private static final String AFDS_PREFIX = "AFDS_";
+    public static final String UPLOADED_AF_SUFFIX = " (uploaded)";
     private static final String UPLOADED_FILES_PATH_PROPERTY_KEY = "org.aksw.gerbil.UploadPath";
     private static final String UPLOADED_DATASET_SUFFIX = " (uploaded)";
     private static final String UPLOADED_DATASET_PREFIX = "NIFDS_";
@@ -164,8 +165,8 @@ public AnnotatorConfiguration getAnnotatorConfig(String name, ExperimentType typ
                 try {
                     return new InstanceListBasedConfigurationImpl(name, false,
                             new DatasetConfigurationImpl(datasetName, false,
-                                    FileBasedQALDDataset.class.getConstructor(String.class, String.class, String.class),
-                                    new Object[] { datasetName, uploadedFilesPath + fileName, questionLanguage},
+                                    FileBasedQALDDataset.class.getConstructor(String.class, String.class, String.class, String.class),
+                                    new Object[] { datasetName, datasetName.replace(AFDS_PREFIX, ""), uploadedFilesPath + fileName, questionLanguage},
                                     ExperimentType.QA, null, null),
                             type, questionLanguage);
                 } catch (Exception e) {
@@ -207,8 +208,9 @@ public DatasetConfiguration getDatasetConfig(String name, ExperimentType type, S
                 }
                 String uri = uploadedFilesPath + name.substring(brackets[0] + 1, brackets[1]);
                 // remove dataset prefix from the name
-                name = name.substring(UPLOADED_DATASET_PREFIX.length(), brackets[0]) + UPLOADED_DATASET_SUFFIX;
-return new QALDFileDatasetConfig(name, uri, false, type, entityCheckerManager, globalRetriever);
+                String questionLabel = name.substring(UPLOADED_DATASET_PREFIX.length(), brackets[0]);
+                name = questionLabel + UPLOADED_DATASET_SUFFIX;
+return new QALDFileDatasetConfig(name, questionLabel, uri, false, type, entityCheckerManager, globalRetriever);
             }
             if (name.startsWith(AF_PREFIX)) {
                 // This describes a QA answer file
@@ -223,6 +225,11 @@ public DatasetConfiguration getDatasetConfig(String name, ExperimentType type, S
                 String datasetName = name.substring(brackets[0] + 1, brackets[1]);
                 return getDatasetConfig(datasetName, type, questionLanguage);
             }
+            if(name.startsWith(AFDS_PREFIX)){
+                String formatted= name.replace(AFDS_PREFIX, "");
+                return getDatasetConfig("NIFDS_"+formatted+"("+formatted+")", type, questionLanguage);
+
+            }
         }
         LOGGER.error("Got an unknown annotator name\"" + name + "\". Returning null.");
         return null;