issue #154: subject indexing analysis

pkiraly · Aug 16, 2022 · d8a0bfe · d8a0bfe
1 parent fbc1126
commit d8a0bfe
Show file tree

Hide file tree

Showing 3 changed files with 38 additions and 12 deletions.
diff --git a/src/main/java/de/gwdg/metadataqa/marc/analysis/ClassificationAnalyzer.java b/src/main/java/de/gwdg/metadataqa/marc/analysis/ClassificationAnalyzer.java
@@ -5,6 +5,7 @@
 import de.gwdg.metadataqa.marc.dao.MarcRecord;
 import de.gwdg.metadataqa.marc.MarcSubfield;
 import de.gwdg.metadataqa.marc.cli.utils.Schema;
+import de.gwdg.metadataqa.marc.definition.bibliographic.SchemaType;
 import de.gwdg.metadataqa.marc.definition.general.indexer.subject.ClassificationSchemes;
 
 import java.util.ArrayList;
@@ -71,14 +72,21 @@ public class ClassificationAnalyzer {
     "653"  // Index Term - Uncontrolled
   );
 
-  private static final List<FieldWithScheme> fieldsWithScheme = Arrays.asList(
+  private static final List<FieldWithScheme> MARC21_FIELD_WITH_SCHEMES = Arrays.asList(
     new FieldWithScheme("080", "Universal Decimal Classification"),
     new FieldWithScheme("082", "Dewey Decimal Classification"),
     new FieldWithScheme("083", "Dewey Decimal Classification"),
     new FieldWithScheme("085", "Dewey Decimal Classification")
     // new FieldWithScheme("086", "Government Document Classification");
   );
 
+  private static final List<FieldWithScheme> PICA_FIELDS_WITH_SCHEME = Arrays.asList(
+    new FieldWithScheme("045A", "Library of Congress Classification"),
+    new FieldWithScheme("045B", "Other scheme"),
+    new FieldWithScheme("045F", "Dewey Decimal Classification"),
+    new FieldWithScheme("045R", "Regensburger Verbundklassifikation")
+  );
+
   public ClassificationAnalyzer(MarcRecord marcRecord, ClassificationStatistics statistics) {
     this.marcRecord = marcRecord;
     this.statistics = statistics;
@@ -88,11 +96,15 @@ public int process() {
     var total = 0;
     schemasInRecord = new ArrayList<>();
 
-    total = processFieldsWithIndicator1AndSubfield2(total);
-    total = processFieldsWithIndicator2AndSubfield2(total);
-    total = processFieldsWithSubfield2(total);
-    total = processFieldsWithoutSource(total);
-    total = processFieldsWithScheme(total);
+    if (marcRecord.getSchemaType().equals(SchemaType.MARC21)) {
+      total = processFieldsWithIndicator1AndSubfield2(total);
+      total = processFieldsWithIndicator2AndSubfield2(total);
+      total = processFieldsWithSubfield2(total);
+      total = processFieldsWithoutSource(total);
+      total = processFieldsWithScheme(total, MARC21_FIELD_WITH_SCHEMES);
+    } else if (marcRecord.getSchemaType().equals(SchemaType.PICA)) {
+      total = processFieldsWithScheme(total, PICA_FIELDS_WITH_SCHEME);
+    }
 
     increaseCounters(total);
 
@@ -114,7 +126,7 @@ private void increaseCounters(int total) {
       count(collocation, statistics.getCollocationHistogram());
   }
 
-  private int processFieldsWithScheme(int total) {
+  private int processFieldsWithScheme(int total, List<FieldWithScheme> fieldsWithScheme) {
     for (FieldWithScheme fieldWithScheme : fieldsWithScheme) {
       var count = processFieldWithScheme(marcRecord, fieldWithScheme);
       if (count > 0)

diff --git a/src/main/java/de/gwdg/metadataqa/marc/dao/MarcRecord.java b/src/main/java/de/gwdg/metadataqa/marc/dao/MarcRecord.java
@@ -36,6 +36,13 @@ public class MarcRecord implements Extractable, Validatable, Serializable {
   private static final Pattern dataFieldPattern = Pattern.compile("^(\\d\\d\\d)\\$(.*)$");
   private static final Pattern positionalPattern = Pattern.compile("^(Leader|00[678])/(.*)$");
   private static final List<String> simpleControlTags = Arrays.asList("001", "003", "005");
+  private static final List<String> MARC21_SUBJECT_TAGS = Arrays.asList(
+    "052", "055", "072", "080", "082", "083", "084", "085", "086",
+    "600", "610", "611", "630", "647", "648", "650", "651",
+    "653", "654", "655", "656", "657", "658", "662"
+    );
+  public static final List<String> PICA_SUBJECT_TAGS = Arrays.asList("045A", "045B", "045F", "045R");
+
   private static final Map<String, Boolean> undefinedTags = new HashMap<>();
 
   private Leader leader;
@@ -745,11 +752,15 @@ public List<DataField> getAuthorityFields() {
 
   public List<DataField> getSubjects() {
     List<DataField> subjects = new ArrayList<>();
-    List<String> tags = Arrays.asList(
-      "052", "055", "072", "080", "082", "083", "084", "085", "086",
-      "600", "610", "611", "630", "647", "648", "650", "651",
-      "653", "654", "655", "656", "657", "658", "662"
-    );
+    List<String> tags;
+    switch (schemaType) {
+      case PICA:
+        tags = PICA_SUBJECT_TAGS; break;
+      case MARC21:
+      default:
+        tags = MARC21_SUBJECT_TAGS; break;
+    }
+    logger.info("tags: " + tags);
     for (String tag : tags) {
       List<DataField> fields = getDatafield(tag);
       if (fields != null && !fields.isEmpty())

diff --git a/...ava/de/gwdg/metadataqa/marc/definition/general/indexer/subject/ClassificationSchemes.java b/...ava/de/gwdg/metadataqa/marc/definition/general/indexer/subject/ClassificationSchemes.java
@@ -54,6 +54,9 @@ private void initialize() {
     schemes.put("Chronological term", "chronological");
     schemes.put("Geographic name", "geographic");
     schemes.put("Genre/form term", "genre");
+
+    // PICA
+    schemes.put("Regensburger Verbundklassifikation", "rvk");
   }
 
   public String resolve(String key) {