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

Changed rules and dictionaries

Browse Source
This commit is contained in:
deiflaender 2020-11-17 09:56:43 +01:00
parent 4a6631f5b8
commit a4f8d2f424
2 changed files with 20 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
redaction-service-v1/redaction-service-server-v1/src/test/resources/dictionaries/hint_only.txt
View File

@ -1,2 +1 @@
guideline
unpublished unpublished

45
redaction-service-v1/redaction-service-server-v1/src/test/resources/drools/rules.drl
View File

@ -9,8 +9,8 @@ rule "1: Redacted because Section contains Vertebrate"
when when
Section(matchesType("vertebrate")) Section(matchesType("vertebrate"))
then then
section.redact("name", 1, "Redacted because Section contains Vertebrate", "Reg (EC) No 1107/2009 Art. 63 (2g)"); section.redact("name", 1, "Vertebrate found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
section.redact("address", 1, "Redacted because Section contains Vertebrate", "Reg (EC) No 1107/2009 Art. 63 (2g)"); section.redact("address", 1, "Vertebrate found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
end end
@ -18,8 +18,8 @@ rule "2: Not Redacted because Section contains no Vertebrate"
when when
Section(!matchesType("vertebrate")) Section(!matchesType("vertebrate"))
then then
section.redactNot("name", 2, "Not Redacted because Section contains no Vertebrate"); section.redactNot("name", 2, "No Vertebrate found");
section.redactNot("address", 2, "Not Redacted because Section contains no Vertebrate"); section.redactNot("address", 2, "No Vertebrate found");
end end
@ -27,26 +27,26 @@ rule "3: Do not redact Names and Addresses if no redaction Indicator is containe
when when
Section(matchesType("vertebrate"), matchesType("no_redaction_indicator")) Section(matchesType("vertebrate"), matchesType("no_redaction_indicator"))
then then
section.redactNot("name", 3, "Vertebrate was found, but also a no redaction indicator"); section.redactNot("name", 3, "Vertebrate and No Redaction Indicator found");
section.redactNot("address", 3, "Vertebrate was found, but also a no redaction indicator"); section.redactNot("address", 3, "Vertebrate and No Redaction Indicator found");
end end
rule "4: Redact Names and Addresses if no_redaction_indicator and redaction_indicator is contained" rule "4: Do not redact Names and Addresses if no redaction Indicator is contained"
when
Section(matchesType("vertebrate"), matchesType("published_information"))
then
section.redactNot("name", 4, "Vertebrate and Published Information found");
section.redactNot("address", 4, "Vertebrate and Published Information found");
end
rule "5: Redact Names and Addresses if no_redaction_indicator and redaction_indicator is contained"
when when
Section(matchesType("vertebrate"), matchesType("no_redaction_indicator"), matchesType("redaction_indicator")) Section(matchesType("vertebrate"), matchesType("no_redaction_indicator"), matchesType("redaction_indicator"))
then then
section.redact("name", 4, "Vertebrate was found and no_redaction_indicator and redaction_indicator", "Reg (EC) No 1107/2009 Art. 63 (2g)"); section.redact("name", 5, "Vertebrate and Redaction Indicator found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
section.redact("address", 4, "Vertebrate was found and no_redaction_indicator and redaction_indicator", "Reg (EC) No 1107/2009 Art. 63 (2g)"); section.redact("address", 5, "Vertebrate and Redaction Indicator found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
end
rule "5: Do not redact in guideline sections"
when
Section(headlineContainsWord("guideline") || headlineContainsWord("Guidance"))
then
section.redactNot("name", 5, "Section is a guideline section.");
section.redactNot("address", 5, "Section is a guideline section.");
end end
@ -54,8 +54,6 @@ rule "6: Redact contact information if applicant is found"
when when
Section(headlineContainsWord("applicant") || text.contains("Applicant") || headlineContainsWord("Primary contact") || headlineContainsWord("Alternative contact")) Section(headlineContainsWord("applicant") || text.contains("Applicant") || headlineContainsWord("Primary contact") || headlineContainsWord("Alternative contact"))
then then
section.redactLineAfter("Name:", "address", 6, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
section.redactBetween("Address:", "Contact", "address", 6, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
section.redactLineAfter("Contact point:", "address", 6, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2g)"); section.redactLineAfter("Contact point:", "address", 6, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
section.redactLineAfter("Phone:", "address", 6, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2g)"); section.redactLineAfter("Phone:", "address", 6, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
section.redactLineAfter("Fax:", "address", 6, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2g)"); section.redactLineAfter("Fax:", "address", 6, "Applicant information was found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
@ -81,11 +79,8 @@ rule "7: Redact contact information if Producer is found"
when when
Section(text.toLowerCase().contains("producer of the plant protection") || text.toLowerCase().contains("producer of the active substance") || text.contains("Manufacturer of the active substance") || text.contains("Manufacturer:") || text.contains("Producer or producers of the active substance")) Section(text.toLowerCase().contains("producer of the plant protection") || text.toLowerCase().contains("producer of the active substance") || text.contains("Manufacturer of the active substance") || text.contains("Manufacturer:") || text.contains("Producer or producers of the active substance"))
then then
section.redactLineAfter("Name:", "address", 7, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2g)"); section.redactLineAfter("Contact:", "address", 7, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
section.redactBetween("Address:", "Contact", "address", 7, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2g)"); section.redactLineAfter("Contact:", "address", 7, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
section.redactBetween("Contact:", "Phone", "address", 7, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
section.redactBetween("Contact:", "Telephone number:", "address", 7, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
section.redactBetween("Address:", "Manufacturing", "address", 7, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
section.redactLineAfter("Telephone:", "address", 7, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2g)"); section.redactLineAfter("Telephone:", "address", 7, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
section.redactLineAfter("Phone:", "address", 7, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2g)"); section.redactLineAfter("Phone:", "address", 7, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2g)");
section.redactLineAfter("Fax:", "address", 7, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2g)"); section.redactLineAfter("Fax:", "address", 7, "Producer was found", "Reg (EC) No 1107/2009 Art. 63 (2g)");