Marmelator/redaction-service
1
0
Fork 0
Code Issues Pull Requests 14 Actions Packages Projects Releases 622 Wiki Activity

RED-7902: fix author recommendation via rules

Browse Source
This commit is contained in:
Kilian Schüttler 2023-11-23 15:22:44 +01:00
parent 5c652df904
commit 74e7a6fb68
5 changed files with 123 additions and 68 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/RedactionIntegrationTest.java
View File

@ -970,7 +970,11 @@ public class RedactionIntegrationTest extends AbstractRedactionIntegrationTest {
redactionLog.getEntityLogEntry().forEach(entry -> {
if (!entry.getEntryType().equals(EntryType.HINT)) {
assertThat(entry.getReason()).isEqualTo("Not redacted because it's row does not belong to a vertebrate study");
if (entry.getType().equals("CBI_author")) {
assertThat(entry.getReason()).isEqualTo("Not redacted because it's row does not belong to a vertebrate study");
} else if (entry.getType().equals("CBI_address")) {
assertThat(entry.getReason()).isEqualTo("No vertebrate found");
}
}
});
}

2
redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/acceptance_rules.drl
View File

@ -218,7 +218,7 @@ rule "CBI.11.0: Recommend all CBI_author entities in Table with Vertebrate Study
when
$table: Table(hasHeader("Author(s)") && hasHeader("Vertebrate Study Y/N"))
then
$table.getEntitiesOfType("CBI_author").forEach(entity -> dictionary.addMultipleAuthorsAsRecommendation(entity));
$table.getEntitiesOfType("CBI_author").stream().filter(IEntity::applied).forEach(entity -> dictionary.addMultipleAuthorsAsRecommendation(entity));
end

61
redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/all_redact_manager_rules.drl
View File

@ -434,42 +434,59 @@ rule "CBI.11.0: Recommend all CBI_author entities in Table with Vertebrate Study
when
$table: Table(hasHeader("Author(s)") && hasHeader("Vertebrate Study Y/N"))
then
$table.getEntitiesOfType("CBI_author").forEach(entity -> dictionary.addMultipleAuthorsAsRecommendation(entity));
$table.getEntitiesOfType("CBI_author").stream().filter(IEntity::applied).forEach(entity -> dictionary.addMultipleAuthorsAsRecommendation(entity));
end
// Rule unit: CBI.12
rule "CBI.12.0: Add all cells with Header Author(s) as CBI_author"
salience 1
rule "CBI.12.0: Redact and recommend TableCell with header 'Author' or 'Author(s)' and header 'Vertebrate study Y/N' with value 'Yes' (non vertebrate study)"
agenda-group "LOCAL_DICTIONARY_ADDS"
when
$table: Table(hasHeader("Author(s)") || hasHeader("Author"))
not FileAttribute(label == "Vertebrate Study", value.toLowerCase() == "yes")
$table: Table(hasHeader("Author(s)") || hasHeader("Author"), hasHeaderIgnoreCase("Vertebrate Study Y/N"))
TableCell(header, containsAnyStringIgnoreCase("Author", "Author(s)"), $authorCol: col) from $table.streamHeaders().toList()
TableCell(header, containsStringIgnoreCase("Vertebrate study Y/N"), $vertebrateCol: col) from $table.streamHeaders().toList()
TableCell(!header, containsAnyString("Yes", "Y"), $rowWithYes: row) from $table.streamCol($vertebrateCol).toList()
$authorCell: TableCell(row == $rowWithYes) from $table.streamCol($authorCol).toList()
then
Stream.concat(
$table.streamTableCellsWithHeader("Author(s)"),
$table.streamTableCellsWithHeader("Author")
)
.map(tableCell -> entityCreationService.bySemanticNode(tableCell, "CBI_author", EntityType.ENTITY))
.filter(Optional::isPresent)
.map(Optional::get)
.forEach(redactionEntity -> redactionEntity.skip("CBI.12.0", "Author(s) header found"));
entityCreationService.bySemanticNode($authorCell, "CBI_author", EntityType.ENTITY)
.ifPresent(authorEntity -> {
authorEntity.redact("CBI.12.0", "Redacted because it's row belongs to a vertebrate study", "Article 39(e)(3) of Regulation (EC) No 178/2002");
dictionary.addMultipleAuthorsAsRecommendation(authorEntity);
});
end
rule "CBI.12.1: Do not redact CBI_author, if its row contains a cell with header \"Vertebrate study Y/N\" and value No"
rule "CBI.12.1: Redact and recommend TableCell with header 'Author' or 'Author(s)' and header 'Vertebrate study Y/N' with value 'Yes' (vertebrate study)"
agenda-group "LOCAL_DICTIONARY_ADDS"
when
$table: Table(hasRowWithHeaderAndValue("Vertebrate study Y/N", "N") || hasRowWithHeaderAndValue("Vertebrate study Y/N", "No"))
FileAttribute(label == "Vertebrate Study", value.toLowerCase() == "yes")
$table: Table(hasHeader("Author(s)") || hasHeader("Author"), hasHeaderIgnoreCase("Vertebrate Study Y/N"))
TableCell(header, containsAnyStringIgnoreCase("Author", "Author(s)"), $authorCol: col) from $table.streamHeaders().toList()
TableCell(header, containsStringIgnoreCase("Vertebrate study Y/N"), $vertebrateCol: col) from $table.streamHeaders().toList()
TableCell(!header, containsAnyString("Yes", "Y"), $rowWithYes: row) from $table.streamCol($vertebrateCol).toList()
$authorCell: TableCell(row == $rowWithYes) from $table.streamCol($authorCol).toList()
then
$table.streamEntitiesWhereRowHasHeaderAndAnyValue("Vertebrate study Y/N", List.of("N", "No"))
.filter(redactionEntity -> redactionEntity.isAnyType(List.of("CBI_author", "CBI_address")))
.forEach(authorEntity -> authorEntity.skip("CBI.12.1", "Not redacted because it's row does not belong to a vertebrate study"));
entityCreationService.bySemanticNode($authorCell, "CBI_author", EntityType.ENTITY)
.ifPresent(authorEntity -> {
authorEntity.redact("CBI.12.1", "Redacted because it's row belongs to a vertebrate study", "Article 39(e)(2) of Regulation (EC) No 178/2002");
dictionary.addMultipleAuthorsAsRecommendation(authorEntity);
});
end
rule "CBI.12.2: Redact CBI_author, if its row contains a cell with header \"Vertebrate study Y/N\" and value Yes"
rule "CBI.12.2: Skip TableCell with header 'Author' or 'Author(s)' and header 'Vertebrate study Y/N' with value 'No'"
when
$table: Table(hasRowWithHeaderAndValue("Vertebrate study Y/N", "Y") || hasRowWithHeaderAndValue("Vertebrate study Y/N", "Yes"))
$table: Table(hasHeader("Author(s)") || hasHeader("Author"), hasHeaderIgnoreCase("Vertebrate Study Y/N"))
TableCell(header, containsAnyStringIgnoreCase("Author", "Author(s)"), $authorCol: col) from $table.streamHeaders().toList()
TableCell(header, containsStringIgnoreCase("Vertebrate study Y/N"), $vertebrateCol: col) from $table.streamHeaders().toList()
TableCell(!header, containsAnyString("No", "N"), $rowWithNo: row) from $table.streamCol($vertebrateCol).toList()
$authorCell: TableCell(row == $rowWithNo) from $table.streamCol($authorCol).toList()
then
$table.streamEntitiesWhereRowHasHeaderAndAnyValue("Vertebrate study Y/N", List.of("Y", "Yes"))
.filter(redactionEntity -> redactionEntity.isAnyType(List.of("CBI_author", "CBI_address")))
.forEach(authorEntity -> authorEntity.redact("CBI.12.2", "Redacted because it's row belongs to a vertebrate study", "Reg (EC) No 1107/2009 Art. 63 (2g)"));
entityCreationService.bySemanticNode($authorCell, "CBI_author", EntityType.ENTITY)
.ifPresent(authorEntity -> authorEntity.skip("CBI.12.2", "Not redacted because it's row does not belong to a vertebrate study"));
end

61
redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/rules.drl
View File

@ -281,42 +281,59 @@ rule "CBI.11.0: Recommend all CBI_author entities in Table with Vertebrate Study
when
$table: Table(hasHeader("Author(s)") && hasHeader("Vertebrate Study Y/N"))
then
$table.getEntitiesOfType("CBI_author").forEach(entity -> dictionary.addMultipleAuthorsAsRecommendation(entity));
$table.getEntitiesOfType("CBI_author").stream().filter(IEntity::applied).forEach(entity -> dictionary.addMultipleAuthorsAsRecommendation(entity));
end
// Rule unit: CBI.12
rule "CBI.12.0: Add all cells with Header Author(s) as CBI_author"
salience 1
rule "CBI.12.0: Redact and recommend TableCell with header 'Author' or 'Author(s)' and header 'Vertebrate study Y/N' with value 'Yes' (non vertebrate study)"
agenda-group "LOCAL_DICTIONARY_ADDS"
when
$table: Table(hasHeader("Author(s)") || hasHeader("Author"))
not FileAttribute(label == "Vertebrate Study", value.toLowerCase() == "yes")
$table: Table(hasHeader("Author(s)") || hasHeader("Author"), hasHeaderIgnoreCase("Vertebrate Study Y/N"))
TableCell(header, containsAnyStringIgnoreCase("Author", "Author(s)"), $authorCol: col) from $table.streamHeaders().toList()
TableCell(header, containsStringIgnoreCase("Vertebrate study Y/N"), $vertebrateCol: col) from $table.streamHeaders().toList()
TableCell(!header, containsAnyString("Yes", "Y"), $rowWithYes: row) from $table.streamCol($vertebrateCol).toList()
$authorCell: TableCell(row == $rowWithYes) from $table.streamCol($authorCol).toList()
then
Stream.concat(
$table.streamTableCellsWithHeader("Author(s)"),
$table.streamTableCellsWithHeader("Author")
)
.map(tableCell -> entityCreationService.bySemanticNode(tableCell, "CBI_author", EntityType.ENTITY))
.filter(Optional::isPresent)
.map(Optional::get)
.forEach(redactionEntity -> redactionEntity.skip("CBI.12.0", "Author(s) header found"));
entityCreationService.bySemanticNode($authorCell, "CBI_author", EntityType.ENTITY)
.ifPresent(authorEntity -> {
authorEntity.redact("CBI.12.0", "Redacted because it's row belongs to a vertebrate study", "Article 39(e)(3) of Regulation (EC) No 178/2002");
dictionary.addMultipleAuthorsAsRecommendation(authorEntity);
});
end
rule "CBI.12.1: Do not redact CBI_author, if its row contains a cell with header \"Vertebrate study Y/N\" and value No"
rule "CBI.12.1: Redact and recommend TableCell with header 'Author' or 'Author(s)' and header 'Vertebrate study Y/N' with value 'Yes' (vertebrate study)"
agenda-group "LOCAL_DICTIONARY_ADDS"
when
$table: Table(hasRowWithHeaderAndValue("Vertebrate study Y/N", "N") || hasRowWithHeaderAndValue("Vertebrate study Y/N", "No"))
FileAttribute(label == "Vertebrate Study", value.toLowerCase() == "yes")
$table: Table(hasHeader("Author(s)") || hasHeader("Author"), hasHeaderIgnoreCase("Vertebrate Study Y/N"))
TableCell(header, containsAnyStringIgnoreCase("Author", "Author(s)"), $authorCol: col) from $table.streamHeaders().toList()
TableCell(header, containsStringIgnoreCase("Vertebrate study Y/N"), $vertebrateCol: col) from $table.streamHeaders().toList()
TableCell(!header, containsAnyString("Yes", "Y"), $rowWithYes: row) from $table.streamCol($vertebrateCol).toList()
$authorCell: TableCell(row == $rowWithYes) from $table.streamCol($authorCol).toList()
then
$table.streamEntitiesWhereRowHasHeaderAndAnyValue("Vertebrate study Y/N", List.of("N", "No"))
.filter(redactionEntity -> redactionEntity.isAnyType(List.of("CBI_author", "CBI_address")))
.forEach(authorEntity -> authorEntity.skip("CBI.12.1", "Not redacted because it's row does not belong to a vertebrate study"));
entityCreationService.bySemanticNode($authorCell, "CBI_author", EntityType.ENTITY)
.ifPresent(authorEntity -> {
authorEntity.redact("CBI.12.1", "Redacted because it's row belongs to a vertebrate study", "Article 39(e)(2) of Regulation (EC) No 178/2002");
dictionary.addMultipleAuthorsAsRecommendation(authorEntity);
});
end
rule "CBI.12.2: Redact CBI_author, if its row contains a cell with header \"Vertebrate study Y/N\" and value Yes"
rule "CBI.12.2: Skip TableCell with header 'Author' or 'Author(s)' and header 'Vertebrate study Y/N' with value 'No'"
when
$table: Table(hasRowWithHeaderAndValue("Vertebrate study Y/N", "Y") || hasRowWithHeaderAndValue("Vertebrate study Y/N", "Yes"))
$table: Table(hasHeader("Author(s)") || hasHeader("Author"), hasHeaderIgnoreCase("Vertebrate Study Y/N"))
TableCell(header, containsAnyStringIgnoreCase("Author", "Author(s)"), $authorCol: col) from $table.streamHeaders().toList()
TableCell(header, containsStringIgnoreCase("Vertebrate study Y/N"), $vertebrateCol: col) from $table.streamHeaders().toList()
TableCell(!header, containsAnyString("No", "N"), $rowWithNo: row) from $table.streamCol($vertebrateCol).toList()
$authorCell: TableCell(row == $rowWithNo) from $table.streamCol($authorCol).toList()
then
$table.streamEntitiesWhereRowHasHeaderAndAnyValue("Vertebrate study Y/N", List.of("Y", "Yes"))
.filter(redactionEntity -> redactionEntity.isAnyType(List.of("CBI_author", "CBI_address")))
.forEach(authorEntity -> authorEntity.redact("CBI.12.2", "Redacted because it's row belongs to a vertebrate study", "Reg (EC) No 1107/2009 Art. 63 (2g)"));
entityCreationService.bySemanticNode($authorCell, "CBI_author", EntityType.ENTITY)
.ifPresent(authorEntity -> authorEntity.skip("CBI.12.2", "Not redacted because it's row does not belong to a vertebrate study"));
end

61
redaction-service-v1/rules-management/src/main/resources/all_redact_manager_rules.drl
View File

@ -434,42 +434,59 @@ rule "CBI.11.0: Recommend all CBI_author entities in Table with Vertebrate Study
when
$table: Table(hasHeader("Author(s)") && hasHeader("Vertebrate Study Y/N"))
then
$table.getEntitiesOfType("CBI_author").forEach(entity -> dictionary.addMultipleAuthorsAsRecommendation(entity));
$table.getEntitiesOfType("CBI_author").stream().filter(IEntity::applied).forEach(entity -> dictionary.addMultipleAuthorsAsRecommendation(entity));
end
// Rule unit: CBI.12
rule "CBI.12.0: Add all cells with Header Author(s) as CBI_author"
salience 1
rule "CBI.12.0: Redact and recommend TableCell with header 'Author' or 'Author(s)' and header 'Vertebrate study Y/N' with value 'Yes' (non vertebrate study)"
agenda-group "LOCAL_DICTIONARY_ADDS"
when
$table: Table(hasHeader("Author(s)") || hasHeader("Author"))
not FileAttribute(label == "Vertebrate Study", value.toLowerCase() == "yes")
$table: Table(hasHeader("Author(s)") || hasHeader("Author"), hasHeaderIgnoreCase("Vertebrate Study Y/N"))
TableCell(header, containsAnyStringIgnoreCase("Author", "Author(s)"), $authorCol: col) from $table.streamHeaders().toList()
TableCell(header, containsStringIgnoreCase("Vertebrate study Y/N"), $vertebrateCol: col) from $table.streamHeaders().toList()
TableCell(!header, containsAnyString("Yes", "Y"), $rowWithYes: row) from $table.streamCol($vertebrateCol).toList()
$authorCell: TableCell(row == $rowWithYes) from $table.streamCol($authorCol).toList()
then
Stream.concat(
$table.streamTableCellsWithHeader("Author(s)"),
$table.streamTableCellsWithHeader("Author")
)
.map(tableCell -> entityCreationService.bySemanticNode(tableCell, "CBI_author", EntityType.ENTITY))
.filter(Optional::isPresent)
.map(Optional::get)
.forEach(redactionEntity -> redactionEntity.skip("CBI.12.0", "Author(s) header found"));
entityCreationService.bySemanticNode($authorCell, "CBI_author", EntityType.ENTITY)
.ifPresent(authorEntity -> {
authorEntity.redact("CBI.12.0", "Redacted because it's row belongs to a vertebrate study", "Article 39(e)(3) of Regulation (EC) No 178/2002");
dictionary.addMultipleAuthorsAsRecommendation(authorEntity);
});
end
rule "CBI.12.1: Do not redact CBI_author, if its row contains a cell with header \"Vertebrate study Y/N\" and value No"
rule "CBI.12.1: Redact and recommend TableCell with header 'Author' or 'Author(s)' and header 'Vertebrate study Y/N' with value 'Yes' (vertebrate study)"
agenda-group "LOCAL_DICTIONARY_ADDS"
when
$table: Table(hasRowWithHeaderAndValue("Vertebrate study Y/N", "N") || hasRowWithHeaderAndValue("Vertebrate study Y/N", "No"))
FileAttribute(label == "Vertebrate Study", value.toLowerCase() == "yes")
$table: Table(hasHeader("Author(s)") || hasHeader("Author"), hasHeaderIgnoreCase("Vertebrate Study Y/N"))
TableCell(header, containsAnyStringIgnoreCase("Author", "Author(s)"), $authorCol: col) from $table.streamHeaders().toList()
TableCell(header, containsStringIgnoreCase("Vertebrate study Y/N"), $vertebrateCol: col) from $table.streamHeaders().toList()
TableCell(!header, containsAnyString("Yes", "Y"), $rowWithYes: row) from $table.streamCol($vertebrateCol).toList()
$authorCell: TableCell(row == $rowWithYes) from $table.streamCol($authorCol).toList()
then
$table.streamEntitiesWhereRowHasHeaderAndAnyValue("Vertebrate study Y/N", List.of("N", "No"))
.filter(redactionEntity -> redactionEntity.isAnyType(List.of("CBI_author", "CBI_address")))
.forEach(authorEntity -> authorEntity.skip("CBI.12.1", "Not redacted because it's row does not belong to a vertebrate study"));
entityCreationService.bySemanticNode($authorCell, "CBI_author", EntityType.ENTITY)
.ifPresent(authorEntity -> {
authorEntity.redact("CBI.12.1", "Redacted because it's row belongs to a vertebrate study", "Article 39(e)(2) of Regulation (EC) No 178/2002");
dictionary.addMultipleAuthorsAsRecommendation(authorEntity);
});
end
rule "CBI.12.2: Redact CBI_author, if its row contains a cell with header \"Vertebrate study Y/N\" and value Yes"
rule "CBI.12.2: Skip TableCell with header 'Author' or 'Author(s)' and header 'Vertebrate study Y/N' with value 'No'"
when
$table: Table(hasRowWithHeaderAndValue("Vertebrate study Y/N", "Y") || hasRowWithHeaderAndValue("Vertebrate study Y/N", "Yes"))
$table: Table(hasHeader("Author(s)") || hasHeader("Author"), hasHeaderIgnoreCase("Vertebrate Study Y/N"))
TableCell(header, containsAnyStringIgnoreCase("Author", "Author(s)"), $authorCol: col) from $table.streamHeaders().toList()
TableCell(header, containsStringIgnoreCase("Vertebrate study Y/N"), $vertebrateCol: col) from $table.streamHeaders().toList()
TableCell(!header, containsAnyString("No", "N"), $rowWithNo: row) from $table.streamCol($vertebrateCol).toList()
$authorCell: TableCell(row == $rowWithNo) from $table.streamCol($authorCol).toList()
then
$table.streamEntitiesWhereRowHasHeaderAndAnyValue("Vertebrate study Y/N", List.of("Y", "Yes"))
.filter(redactionEntity -> redactionEntity.isAnyType(List.of("CBI_author", "CBI_address")))
.forEach(authorEntity -> authorEntity.redact("CBI.12.2", "Redacted because it's row belongs to a vertebrate study", "Reg (EC) No 1107/2009 Art. 63 (2g)"));
entityCreationService.bySemanticNode($authorCell, "CBI_author", EntityType.ENTITY)
.ifPresent(authorEntity -> authorEntity.skip("CBI.12.2", "Not redacted because it's row does not belong to a vertebrate study"));
end