Merge pull request #263 from crocs-muni/feat/cert-id-eval

Add ground truth analysis of cert ids.

9 files changed, 486 insertions, 341 deletions

diff --git a/data/cert_id_eval/duplicate_ids.csv b/data/cert_id_eval/duplicate_ids.csv
index 400583bf..4e08e6a8 100644
--- a/data/cert_id_eval/duplicate_ids.csv
+++ b/data/cert_id_eval/duplicate_ids.csv
@@ -1,67 +1,13 @@
 cert_id,id,result,true_id,reason,fixable
-ANSSI-CC-2012/70,1b20b1629ba7d3a6,fp,KECS-ISIS-0490-2014,number of mentions of component certificate higher than actual,scheme mismatch
-ANSSI-CC-2012/70,cb203cf5d91b1ae3,fp,KECS-ISIS-0533-2014,number of mentions of component certificate higher than actual,scheme mismatch
-ANSSI-CC-2012/70,8870ddf70bf57890,fp,KECS-ISIS-0513-2014,number of mentions of component certificate higher than actual,scheme mismatch
-ANSSI-CC-2012/70,3ea02a62856c752a,fp,KECS-ISIS-0468-2013,number of mentions of component certificate higher than actual,scheme mismatch
-ANSSI-CC-2012/70,259e0ab1c4b690cf,fp,KECS-ISIS-0456-2013,number of mentions of component certificate higher than actual,scheme mismatch
-ANSSI-CC-2012/70,aabd0ff0d82fff72,tp,ANSSI-CC-2012/70,,
-ANSSI-CC-2012/70,bcac7e3f9f614ec6,fp,KECS-ISIS-0675-2015,number of mentions of component certificate higher than actual,scheme mismatch
-ANSSI-CC-2012/70,f2327302df1aa660,fp,KECS-ISIS-0435-2013,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-0645-2010,a60b819509d62b84,tp,BSI-DSZ-CC-0645-2010,,
-BSI-DSZ-CC-0645-2010,720aab4f782f4c60,fp,2011-14-INF-1095,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-0645-2010,9e7c513fdd35df17,fp,2011-13-INF-1092,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-0645-2010,2fb4fdf2a570c366,fp,2011-15-INF-1098,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-0645-2010,a5e8267322d0d7a6,fp,2011-12-INF-1089,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-1110-V3-2020,5346d5a4418466ab,fp,KECS-ISIS-1031-2020,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-1110-V3-2020,a7840a951b1de6c2,fp,ANSSI-CC-2020/87,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-1110-V3-2020,09220515a7edf3f9,tp,BSI-DSZ-CC-1110-V3-2020,,
-BSI-DSZ-CC-1110-V3-2020,8b7145c13f36c324,fp,ANSSI-CC-2020/88,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-0782-V5-2020,15d68159595eae09,tp,BSI-DSZ-CC-0782-V5-2020,,
-BSI-DSZ-CC-0782-V5-2020,c8ff17063166050d,fp,KECS-ISIS-1140-2021,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-0782-V5-2020,f2e7c6dfd6a431c8,fp,KECS-ISIS-1139-2021,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-0782-V5-2020,17a23970b35a4f44,fp,KECS-ISIS-1113-2021,number of mentions of component certificate higher than actual,scheme mismatch
 SERTIT-014,a5fd51390f814a46,tp,SERTIT-014,,
 SERTIT-014,8278a9ba639a08d0,fp,,wrong certification report uploaded + clone,
 SERTIT-014,3ba740ca6ee8ed75,fp,,wrong certification report uploaded + clone,
-ANSSI-CC-2021/50,425f13fb7e94add5,tp,ANSSI-CC-2021/50,,
-ANSSI-CC-2021/50,b42009eb34746731,fp,ANSSI-CC-2022/29,number of mentions of component certificate higher than actual,frontpage match + file metadata
-ANSSI-CC-2021/50,0d2cab8061464438,fp,ANSSI-CC-2022/25,number of mentions of component certificate higher than actual,frontpage match + file metadata
-ANSSI-CC-2014/58,27d11629261d8806,fp,KECS-ISIS-0798-2017,number of mentions of component certificate higher than actual,scheme mismatch
-ANSSI-CC-2014/58,627bfd69c2a831cf,tp,ANSSI-CC-2014/58,,
-ANSSI-CC-2014/58,cacdfdafce47678c,fp,KECS-ISIS-0799-2017,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-0223-2003,4408217c4ce524fd,fp,DCSSI-CC-2003/12,number of mentions of component certificate higher than actual + clone,scheme mismatch
-BSI-DSZ-CC-0223-2003,b82b48b0e915aaac,tp,BSI-DSZ-CC-0223-2003,,
-BSI-DSZ-CC-0223-2003,c3f8e8ddb1dabb02,fp,DCSSI-CC-2003/12,number of mentions of component certificate higher than actual + clone,scheme mismatch
-ANSSI-CC-2019/01,f4b4b026e87b5086,fp,KECS-ISIS-0937-2019,number of mentions of component certificate higher than actual,scheme mismatch
-ANSSI-CC-2019/01,fbc78e2407910a4b,fp,KECS-ISIS-0936-2019,number of mentions of component certificate higher than actual,scheme mismatch
-ANSSI-CC-2019/01,84c8bf1f809380bc,tp,ANSSI-CC-2019/01,,
-ANSSI-CC-2018/12,a0c38b4389ad7cf7,fp,2017-63-INF-3232,number of mentions of component certificate higher than actual,scheme mismatch
-ANSSI-CC-2018/12,10107b83e7393c50,tp,ANSSI-CC-2018/12,,
-ANSSI-CC-2018/12,7dfcefd1ffeaa941,fp,2017-62-INF-3233,number of mentions of component certificate higher than actual,scheme mismatch
-ANSSI-CC-2021/49,6fd73557577cf9ed,tp,ANSSI-CC-2021/49,,
-ANSSI-CC-2021/49,037577fc2019fcfa,fp,ANSSI-CC-2022/28,number of mentions of component certificate higher than actual,frontpage match + file metadata
-ANSSI-CC-2021/49,d55eb8c97d71e843,fp,ANSSI-CC-2022/24,number of mentions of component certificate higher than actual,frontpage match + file metadata
 ANSSI-CC-2009/36,fad651102525c5a4,fp,ANSSI-CC-2009/48,mismatched frontpage and title page + clone,
 ANSSI-CC-2009/36,928e364f71d64595,fp,ANSSI-CC-2009/48,mismatched frontpage and title page + clone,
 ANSSI-CC-2009/36,691b730a7bbb28b0,tp,ANSSI-CC-2009/36,,
-567-LSS,0a06ae047b9790ba,tp,567-LSS,clone + removed,handle removed
-567-LSS,53231c8ddc542543,tp,567-LSS,clone,
-567-LSS,3af72451867b3187,tp,567-LSS,clone + removed,handle removed
 CCEVS-VR-VID11024-2019,dfa12d7893c9efdf,fp,,wrong certification report and security target uploaded,
 CCEVS-VR-VID11024-2019,20f851314c4f57d9,tp,CCEVS-VR-VID11024-2019,,
 CCEVS-VR-VID11024-2019,b4288d36bc129d2f,fp,,wrong certification report and security target uploaded,
-BSI-DSZ-CC-0782-V2-2015,188d3cda4e6f2d5d,fp,KECS-ISIS-0677-2015,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-0782-V2-2015,2a91f42389fda90d,fp,KECS-ISIS-0676-2015,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-0782-V2-2015,446aa68e0c4c5083,tp,BSI-DSZ-CC-0782-V2-2015,,
-ANSSI-CC-2021/52,d355c778fcfee3a9,fp,ANSSI-CC-2022/27,number of mentions of component certificate higher than actual,frontpage match + file metadata
-ANSSI-CC-2021/52,6cef8442092da1da,fp,ANSSI-CC-2022/31,number of mentions of component certificate higher than actual,frontpage match + file metadata
-ANSSI-CC-2021/52,536aac44b608f951,tp,ANSSI-CC-2021/52,,
-ANSSI-CC-2021/51,a47713a44e4c656b,tp,ANSSI-CC-2021/51,,
-ANSSI-CC-2021/51,940a18f62c79ffae,fp,ANSSI-CC-2022/30,number of mentions of component certificate higher than actual,frontpage match + file metadata
-ANSSI-CC-2021/51,5787ff8207fd751d,fp,ANSSI-CC-2022/26,number of mentions of component certificate higher than actual,frontpage match + file metadata
-BSI-DSZ-CC-0945-2017,457fcec1ad31841f,fp,ANSSI-CC-2022/37v2,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-0945-2017,0b36924a005c2ae2,fp,ANSSI-CC-2022/36v2,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-0945-2017,5103648897ba1b3a,tp,BSI-DSZ-CC-0945-2017,,
 CCEVS-VR-05-0091,c79dafc7b0da855e,tp,CCEVS-VR-05-0091,clone,
 CCEVS-VR-05-0091,355a5b9240e32cc9,tp,CCEVS-VR-05-0091,clone,
 2009-27-INF-754,7ac69e82d1a680bb,fp,,different reports share cert_id,
@@ -70,16 +16,12 @@ ANSSI-CC-2021/16,0ef56cba2362e463,tp,ANSSI-CC-2021/16,,
 ANSSI-CC-2021/16,875347e0d906197d,fp,ANSSI-CC-2021/16-R01,regex issue,regex fix
 ANSSI-CC-2011/72,2b0342d5b70b7a78,fp,ANSSI-CC-2011/82,mismatched frontpage and title page,
 ANSSI-CC-2011/72,dec61898ab005e45,tp,ANSSI-CC-2011/72,,
-ANSSI-CC-2021/34,60f31793e1e5f4e1,tp,ANSSI-CC-2021/34,,
-ANSSI-CC-2021/34,3cbdd57de3f32ad0,fp,ANSSI-CC-2021/34-R01,regex issue,regex fix
 CCEVS-VR-10786-2017,d976861ecbeee69c,fp,,different reports share cert_id,
 CCEVS-VR-10786-2017,371abb3484840317,fp,,different reports share cert_id,
 ANSSI-CC-2020/65,d492f0e3e19d32f6,fp,ANSSI-CC-2020/65v2,regex substring,frontpage match
 ANSSI-CC-2020/65,b30ef2c1aaec9236,tp,ANSSI-CC-2020/65,,
 BSI-DSZ-CC-0588-2009,03b4ff588fb88dae,fp,,wrong certification report ,
 BSI-DSZ-CC-0588-2009,031667f4e242da61,tp,BSI-DSZ-CC-0588-2009,,
-ANSSI-CC-2021/62,559e4549a852aab2,tp,ANSSI-CC-2021/62,,
-ANSSI-CC-2021/62,3e020408537aaad0,fp,ANSSI-CC-2021/62-R01,regex issue,regex fix
 ANSSI-CC-2020/12,6c6c5ba82977609b,fp,ANSSI-CC-2020/93,number of mentions of component certificate higher than actual,frontpage match
 ANSSI-CC-2020/12,af6767c887ae930a,tp,ANSSI-CC-2020/12,,
 BSI-DSZ-CC-1139-V3-2021,887139711564fdb3,tp,BSI-DSZ-CC-1139-V3-2021,clone,
@@ -102,38 +44,22 @@ CCEVS-VR-10425-2011,e2f3a1cdd592ab05,tp,CCEVS-VR-10425-2011,clone,
 CCEVS-VR-10425-2011,52f07fb00a81780e,tp,CCEVS-VR-10425-2011,clone,
 CCEVS-VR-VID10510-2014,965f41d028746f9b,fp,,different reports share cert_id,
 CCEVS-VR-VID10510-2014,0cf98690ba05ef68,fp,,different reports share cert_id,
-383-4-263-CR,f2032ab2a50ab35d,tp,383-4-263-CR,clone,
-383-4-263-CR,1fb95703e4e12a49,tp,383-4-263-CR,clone,
-NSCIB-CC-0237694-CR,9fd2cc58ecfab9c4,tp,NSCIB-CC-0237694-CR,clone,
-NSCIB-CC-0237694-CR,b1401488f998c1f0,tp,NSCIB-CC-0237694-CR,clone,
-ANSSI-CC-2020/07,8820852f4a1163b3,tp,ANSSI-CC-2020/07,clone,
-ANSSI-CC-2020/07,9ab449ecd0da1754,tp,ANSSI-CC-2020/07,clone,
+383-4-263-CR,f2032ab2a50ab35d,tp,383-4-263,clone,
+383-4-263-CR,1fb95703e4e12a49,tp,383-4-263,clone,
 CCEVS-VR-05-0124,8bf284d3b482ad65,tp,CCEVS-VR-05-0124,clone,
 CCEVS-VR-05-0124,3ad601a9cb30dfef,tp,CCEVS-VR-05-0124,clone,
 CCEVS-VR-05-0102,a67fe4216fd88917,tp,CCEVS-VR-05-0102,clone,
 CCEVS-VR-05-0102,ec25be3458d728cd,tp,CCEVS-VR-05-0102,clone,
-383-4-198-CR,3c24bbe7724dbfdd,tp,383-4-198-CR,clone,
-383-4-198-CR,0a3793d207c47790,tp,383-4-198-CR,clone,
-NSCIB-CC-0308282-CR,01fa8bfbc95c23b5,tp,NSCIB-CC-0308282-CR,clone,
-NSCIB-CC-0308282-CR,f6e744b5e5ae6d4f,tp,NSCIB-CC-0308282-CR,clone + removed,handle removed
-ANSSI-CC-2014/59,ac4687f4b15507b6,fp,KECS-ISIS-0797-2017,number of mentions of component certificate higher than actual,scheme mismatch
-ANSSI-CC-2014/59,6d41460480570d22,tp,ANSSI-CC-2014/59,,
+383-4-198-CR,3c24bbe7724dbfdd,tp,383-4-198,clone,
+383-4-198-CR,0a3793d207c47790,tp,383-4-198,clone,
 CCEVS-VR-VID10327-2009,c80e1ca4d8641a3d,tp,CCEVS-VR-VID10327-2009,clone,
 CCEVS-VR-VID10327-2009,0f6c8bde7c6574e3,tp,CCEVS-VR-VID10327-2009,clone,
-NSCIB-CC-0388535-CR,3a89942886178cad,tp,NSCIB-CC-0388535-CR,clone + removed,handle removed
-NSCIB-CC-0388535-CR,ab4814d2b8c742a0,tp,NSCIB-CC-0388535-CR,clone,
 ANSSI-CC-2010/36,5e4c757231135ba1,tp,ANSSI-CC-2010/36,clone,
 ANSSI-CC-2010/36,69c75283a4ccf934,tp,ANSSI-CC-2010/36,clone,
-BSI-DSZ-CC-0950-V2-2018,a46522d0a592c1fe,tp,BSI-DSZ-CC-0950-V2-2018,clone + removed,handle removed
-BSI-DSZ-CC-0950-V2-2018,6f90b0bf6bbe884b,tp,BSI-DSZ-CC-0950-V2-2018,clone,
-Rapport de certification 2005/01,f3bb1641b4bb0aa0,tp,Rapport de certification 2005/01,clone,
-Rapport de certification 2005/01,749fae44017016c5,tp,Rapport de certification 2005/01,clone + mismatched scheme,scheme mismatch
 2004-3-INF-71,1bb3da30bc8f4750,tp,2004-3-INF-71,clone,
 2004-3-INF-71,0b0648a9b9a26747,tp,2004-3-INF-71,clone,
 CCEVS-VR-VID10559-2014,aa49720eb88f291c,tp,CCEVS-VR-VID10559-2014,,
 CCEVS-VR-VID10559-2014,449ea70a6342d680,fp,,wrong certification report,
-ANSSI-CC-2021/30,c62b878b1ac43df9,tp,ANSSI-CC-2021/30,,
-ANSSI-CC-2021/30,459539da9315ccac,fp,ANSSI-CC-2021/30-R01,regex issue,regex fix
 ANSSI-CC-2010/24,0372b733c28f305a,tp,ANSSI-CC-2010/24,clone,
 ANSSI-CC-2010/24,b9789028f97490ce,tp,ANSSI-CC-2010/24,clone,
 ANSSI-CC-2010/34,2f53e1d94ca1a010,tp,ANSSI-CC-2010/34,clone,
@@ -148,20 +74,12 @@ ANSSI-CC-2014/36,8d6bca0d7f228191,tp,ANSSI-CC-2014/36,clone,
 ANSSI-CC-2014/36,f43765c81befb944,tp,ANSSI-CC-2014/36,clone,
 CCEVS-VR-VID10003-2008,bf7b3c78598f74cd,tp,CCEVS-VR-VID10003-2008,,
 CCEVS-VR-VID10003-2008,8f538d2eb9a1c6c7,fp,,wrong certification report,
-BSI-DSZ-CC-0891-V2-2016,bc715a45d7584d95,fp,NSCIB-CC-16-95781-CR,number of mentions of component certificate higher than actual,frontpage match + scheme mismatch
-BSI-DSZ-CC-0891-V2-2016,2d835bff00514238,tp,BSI-DSZ-CC-0891-V2-2016,,
-BSI-DSZ-CC-0935-2015,025ccea9d90154b0,tp,BSI-DSZ-CC-0935-2015,,
-BSI-DSZ-CC-0935-2015,a4bbf07046109fc9,fp,NSCIB-CC-10-30076-CR,number of mentions of component certificate higher than actual,scheme mismatch
 ANSSI-CC-2009/46,baead73ecc6e4a33,tp,ANSSI-CC-2009/46,clone,
 ANSSI-CC-2009/46,d76f210e41d5589c,tp,ANSSI-CC-2009/46,clone,
-BSI-DSZ-CC-0234-2004,f577607298c4b574,fp,CERTIFICATION REPORT No. P212,number of mentions of component certificate higher than actual,scheme mismatch
-BSI-DSZ-CC-0234-2004,1cb63a13bf4c58ff,tp,BSI-DSZ-CC-0234-2004,,
 BSI-DSZ-CC-1053-2018,4e246ff7fd346630,tp,BSI-DSZ-CC-1053-2018,clone,
 BSI-DSZ-CC-1053-2018,a6629b6ce5514a43,tp,BSI-DSZ-CC-1053-2018,clone,
 ANSSI-CC-2018/41,10b17081dd7cad8f,tp,ANSSI-CC-2018/41,clone,
 ANSSI-CC-2018/41,27baf39406deb767,tp,ANSSI-CC-2018/41,clone,
-ANSSI-CC-2021/02,82c717a9a379b75c,tp,ANSSI-CC-2021/02,,
-ANSSI-CC-2021/02,7e24168630d5124e,fp,ANSSI-CC-2021/02-R01,regex issue,regex fix
 ANSSI-CC-2009/49,1b5a04adf5047f67,tp,ANSSI-CC-2009/49,clone,
 ANSSI-CC-2009/49,1188aebd8681e4fe,tp,ANSSI-CC-2009/49,clone,
 CCEVS-VR-05-0137,feeecec3ac6f0503,tp,CCEVS-VR-05-0137,clone,
@@ -170,26 +88,14 @@ ANSSI-CC-2013/09,a5adb726852f5cc5,tp,ANSSI-CC-2013/09,clone,
 ANSSI-CC-2013/09,ee35dcafbe1b9532,tp,ANSSI-CC-2013/09,clone,
 ANSSI-CC-2010/03,67a79bfc84b4071b,tp,ANSSI-CC-2010/03,clone,
 ANSSI-CC-2010/03,8826a8ae0369901a,tp,ANSSI-CC-2010/03,clone,
-NSCIB-CC-98209-CR5,af86594cd7aaf6d4,tp,NSCIB-CC-98209-CR5,clone + removed,handle removed
-NSCIB-CC-98209-CR5,6fc6a0a8163fea0c,tp,NSCIB-CC-98209-CR5,clone,
-BSI-DSZ-CC-0501-2008,78bb780793fbb176,tp,BSI-DSZ-CC-0501-2008,,
-BSI-DSZ-CC-0501-2008,f7cbb33cc639b5b5,fp,KECS-ISIS-0222-2010,number of mentions of component certificate higher than actual,scheme mismatch
 BSI-DSZ-CC-0196-2003,063d4ed52b084441,fp,,wrong certification report ,
 BSI-DSZ-CC-0196-2003,41708ba21f2bb4a7,tp,BSI-DSZ-CC-0196-2003,,
-BSI-DSZ-CC-0624-2010,09be49631855ae19,tp,BSI-DSZ-CC-0624-2010,,
-BSI-DSZ-CC-0624-2010,1816ddecd7432cf9,fp,KECS-ISIS-0378-2012,number of mentions of component certificate higher than actual,
-383-4-100-CR,8ba5d4d02ea73d3d,tp,383-4-100-CR,,scheme mismatch
+383-4-100-CR,8ba5d4d02ea73d3d,tp,383-4-100,,scheme mismatch
 383-4-100-CR,c4e1391e8ba34694,fp,,wrong certification report ,
-NSCIB-CC-0441513-CR,7533a87b0379ce3e,tp,NSCIB-CC-0441513-CR,clone + removed,handle removed
-NSCIB-CC-0441513-CR,82c6207791448e0c,tp,NSCIB-CC-0441513-CR,clone,
 ANSSI-CC-2019/20,c36783a79de002c5,tp,ANSSI-CC-2019/20,clone,
 ANSSI-CC-2019/20,088504703b9eedc1,tp,ANSSI-CC-2019/20,clone,
-CSEC2020024,f2b4285a59a296f2,tp,CSEC2020024,clone + removed,handle removed
-CSEC2020024,e502549eb7fd9869,tp,CSEC2020024,clone,
 CCEVS-VR-10610-2015,6f0a50d3bd7db7c7,fp,,wrong certification report ,
 CCEVS-VR-10610-2015,e93207ee36a5fabf,tp,CCEVS-VR-10610-2015,,
-NSCIB-CC-0237686-CR,294ef6cb28408ec9,tp,NSCIB-CC-0237686-CR,clone,
-NSCIB-CC-0237686-CR,6bca140482214182,tp,NSCIB-CC-0237686-CR,clone + removed,handle removed
 CCEVS-VR-04-0074,d42974d5cce56320,tp,CCEVS-VR-04-0074,clone,
 CCEVS-VR-04-0074,1bcac3b368582ae1,tp,CCEVS-VR-04-0074,clone,
 ANSSI-CC-2021/37,5498bfcecf19dfa3,tp,ANSSI-CC-2021/37,,
@@ -198,27 +104,37 @@ CCEVS-VR-06-0024,686005d0b5ff5c5c,tp,CCEVS-VR-06-0024,clone,
 CCEVS-VR-06-0024,b901bbd0ce1f0a3b,tp,CCEVS-VR-06-0024,clone,
 ANSSI-CC-2011/07,41bd2924e9afced5,tp,ANSSI-CC-2011/07,clone,
 ANSSI-CC-2011/07,796445972ef2ff5b,tp,ANSSI-CC-2011/07,clone,
-ANSSI-CC-2018/38,3459c14186e8223f,tp,ANSSI-CC-2018/38,,
-ANSSI-CC-2018/38,371dca18821f7714,fp,ANSSI-CC-2018/38v2,regex substring,frontpage match
-CRP292,a00e93f2981ef418,fp,CRP298,number of mentions of component certificate higher than actual,frontpage match
-CRP292,3e345d3c6002ad55,tp,CRP292,,
-ANSSI-2009/24,a52f807db0eb7fdb,tp,ANSSI-2009/24,clone,
-ANSSI-2009/24,e6a8cbbc8c557bed,tp,ANSSI-2009/24,clone,
-ANSSI-CC-2013/47,34370b67b5e675c3,fp,KECS-ISIS-0489-2014,number of mentions of component certificate higher than actual,scheme mismatch
-ANSSI-CC-2013/47,6aa2f56da553d95d,tp,ANSSI-CC-2013/47,,
-ANSSI-CC-2010/02,7d4585a4b5b6e873,fp,KECS-ISIS-0419-2012,number of mentions of component certificate higher than actual,scheme mismatch
-ANSSI-CC-2010/02,dbbf02a1cd0ad33b,tp,ANSSI-CC-2010/02,,
+ANSSI-2009/24,a52f807db0eb7fdb,tp,ANSSI-CC-2009/24,clone,
+ANSSI-2009/24,e6a8cbbc8c557bed,tp,ANSSI-CC-2009/24,clone,
 2016-47-INF-2126,f254bdb3e177193d,tp,2016-47-INF-2126,clone,
 2016-47-INF-2126,b1f92dfbc5d51547,tp,2016-47-INF-2126,clone,
 ANSSI-CC-2010/39,4e9bb649bce14c86,tp,ANSSI-CC-2010/39,clone,
 ANSSI-CC-2010/39,30ae700d801bd41c,tp,ANSSI-CC-2010/39,clone,
 ANSSI-CC-2014/46,da7fa774f888542b,tp,ANSSI-CC-2014/46,clone,
 ANSSI-CC-2014/46,ba5cbf3c600d99ac,tp,ANSSI-CC-2014/46,clone,
-ANSSI-CC-2021/35,14b5a3a574c90154,fp,ANSSI-CC-2021/35-R01,regex issue,regex fix
-ANSSI-CC-2021/35,ce22219c2049985c,tp,ANSSI-CC-2021/35,,
 CSEC2016007,13e306eb69c8bf95,tp,CSEC2016007,clone (recertification?),
 CSEC2016007,0c3ff6669dcf6968,tp,CSEC2016007,clone (recertification?),
-BSI-DSZ-CC-0639-2010,2f9feaf4121720da,tp,BSI-DSZ-CC-0639-2010,,
-BSI-DSZ-CC-0639-2010,db7627e56b8621b1,fp,KECS-ISIS-0394-2012,number of mentions of component certificate higher than actual,scheme mismatch
-Certification Report 98/94,28bb2a8869397c4c,fp,CRP106,number of mentions of component certificate higher than actual,scheme mismatch
-Certification Report 98/94,28a9fc2137d72ce9,fp,CRP103,number of mentions of component certificate higher than actual,scheme mismatch
+CSEC2015006,a5607124c92c10ed,tp,CSEC2015006,shared cert_id,
+CSEC2015006,f89c3a438204a45b,tp,CSEC2015006,shared cert_id,
+ANSSI-CC-2009/47,80ae5f71b5fa278a,tp,ANSSI-CC-2009/47,clone,
+ANSSI-CC-2009/47,0cbcee9f1cde47a8,tp,ANSSI-CC-2009/47,clone,
+CRP141,e510cd16d4f9f4a3,tp,CRP141,clone,
+CRP141,b802ce16d43a545e,tp,CRP141,clone,
+ANSSI-CC-2009/25,fb40bf81f6774748,tp,ANSSI-CC-2009/25,,
+ANSSI-CC-2009/25,b54e179209c533c5,fp,ANSSI-CC-2009/57,mismatched frontpage and title page,
+ANSSI-CC-2018/57,cacd9550c69df042,tp,ANSSI-CC-2018/57,,
+ANSSI-CC-2018/57,449c74a92ebb61a4,fp,ANSSI-CC-2018/57v2,number of mentions of component certificate higher than actual,frontpage match
+CRP153,20c98a49ea86a519,tp,CRP153,clone,
+CRP153,a79f15874e255007,tp,CRP153,clone,
+ANSSI-CC-2018/32,82ade52e9ef12fd9,tp,ANSSI-CC-2018/32,,
+ANSSI-CC-2018/32,09b1192474bc8b27,fp,ANSSI-CC-2018/32v2,number of mentions of component certificate higher than actual,frontpage match
+383-4-335,3caad2930e3cbf2a,tp,383-4-335,,
+383-4-335,97a292f85b058e0d,fp,383-4-435,cert_id typo,
+KECS-NISS-0052-2006,59cbd48b9dc51e87,fp,KECS-NISS-0053-2006,filename typo,
+KECS-NISS-0052-2006,06f6d7907510c36e,tp,KECS-NISS-0052-2006,,
+ANSSI-CC-2018/58,744a7a202d909323,fp,ANSSI-CC-2018/58v2,number of mentions of component certificate higher than actual,frontpage match
+ANSSI-CC-2018/58,d57c84efd02c4bb7,tp,ANSSI-CC-2018/58,,
+CCEVS-VR-03-0044,814a2ac0aeae08b1,tp,CCEVS-VR-03-0044,clone,
+CCEVS-VR-03-0044,02a31d0ed42e4ff1,tp,CCEVS-VR-03-0044,clone,
+ANSSI-CC-2018/56,6df8f86449e56c05,tp,ANSSI-CC-2018/56,,
+ANSSI-CC-2018/56,c5e25f90c7006546,fp,ANSSI-CC-2018/56v2,number of mentions of component certificate higher than actual,frontpage match
\ No newline at end of file
diff --git a/data/cert_id_eval/missing_ids.csv b/data/cert_id_eval/missing_ids.csv
index f7296cbc..d9c38f5e 100644
--- a/data/cert_id_eval/missing_ids.csv
+++ b/data/cert_id_eval/missing_ids.csv
@@ -1,91 +1,80 @@
 id,scheme,cert_id,reason
-889693d2b11bdfaf,JP,Certification No. C0023,not parsed due to PDF issue
-20c98a49ea86a519,UK,,maybe "COMMON CRITERIA CERTIFICATION REPORT NO. P153"
+889693d2b11bdfaf,JP,C0023,pdf issue
 a90d78924e91dc6c,FR,,404 for report PDF
-21d5f1b05e1ba491,JP,Certification No. C0003,not parsed due to PDF issue
+21d5f1b05e1ba491,JP,C0003,pdf issue
 6af4a27251001715,US,,404 for report PDF
-3ad7749ef678ce95,JP,Certification No. C0093,not parsed due to PDF issue
+3ad7749ef678ce95,JP,C0093,pdf issue
 d0d0dfe4860db6be,US,,404 for report PDF
 822e5f25e7c4631e,CA,,cert_id missing
-10dc636e9da3e1b6,CA,,cert_id missing
-357b38a376d955f7,JP,,has security target uploaded in place of report
-e2ecf0daf73716f2,CA,511-LSS,regex missing
+357b38a376d955f7,JP,C0678,has security target uploaded in place of report
 7d074a56ee50aec4,US,,404 for report PDF
 e27eb9efbd67c4f7,KR,,404 for report PDF
 811ca609aebe2e96,US,,404 for report PDF
-9219f2fb5f305269,KR,KECS-ISIS-0119-2008,has unicode hyphens 8208
-10339846e519597c,CA,516-LSS,regex missing
 4a1fa75170579066,US,,cert_id missing
-f0068ddc5569b772,CA,518-LSS,regex missing
-28fb117adbb809e3,SE,CSEC2006002,space in cert_id
-2d05e976e1adf696,CA,,report not a PDF (docx) also cert_id missing
-a79f15874e255007,UK,,maybe "COMMON CRITERIA CERTIFICATION REPORT NO. P153"
-7b1f816f20381626,US,,maybe "10-1904-R-0030 V1.1" the ETR id
-32d4bf275131aa15,KR,KECS-ISIS-0118-2008,has unicode hyphens 8208
+7b1f816f20381626,US,,cert_id missing
 f4dc9b09aa682008,US,,cert_id missing
 ed0b7793ac82822e,US,,cert_id missing
-bfc86f14e2484335,CA,,report not a PDF (docx) also cert_id missing
-1491021aa321b828,CA,517 CCS 2020,regex missing
-b762fcb2983aa148,JP,Certification No. C0046,not parsed due to PDF issue
+b762fcb2983aa148,JP,C0046,pdf issue
 e133881d7203a6e4,ES,INF-3516,regex missing
 305af0eec85bb856,KR,,404 for report PDF
-a91b4c64329d05a8,SE,CSEC2019010,space in cert_id
-c5bad5b309469e22,CA,519-EWA,regex missing
 08e892fc32b0579f,ES,INF-3483,regex missing
-876cb2fe02017f3f,JP,Certification No. C0012,not parsed due to PDF issue
-0dcc691ea04d1b52,CA,,cert_id missing
-f9c8da9deff77ab5,CA,,report not a PDF (docx) also cert_id missing
+0dcc691ea04d1b52,CA,383-4-351,certificate.docx
 ace7ea6d7f58dbb1,DE,,404 for report PDF
 7147c2f70d983d57,TR,21.0.03/TSE-CCCS-44,report not a PDF (docx)
-6266d99d3b983091,CA,,cert_id missing
-66838f2e4bb6eccc,FR,,has security target uploaded in place of report
-8ed77a6223a94fb5,US,,maybe "09-1832-R-0027 V1.0" the ETR id
+66838f2e4bb6eccc,FR,ANSSI-CC-2018/42,has security target uploaded in place of report
+8ed77a6223a94fb5,US,,cert_id missing
 422dc5758723c7d1,TR,21.0.03/TSE-CCCS-68,report not a PDF (docx)
 02482228eb547c15,US,,404 for report PDF
-4de4a9f436958574,CA,525 LSS 2020,report not a PDF (docx)
-c312bbf9e303b844,SE,CSEC2019012,space in cert_id
-fb6d7faf0ed2dca5,CA,,cert_id missing
-b3ae94eeccc566ba,CA,383-4-17-CR,regex missing
 d3568613c552f9e8,KR,,malformed PDF
 b43100f226a61f53,FR,,malformed PDF
-0a117ebd82bdfb0c,US,,maybe "11-2280-R-0077 V1.1" the ETR id
+0a117ebd82bdfb0c,US,,cert_id missing
 95948dc5f3e472fb,IT,,cert_id missing
-9253ec0d31264194,US,,maybe "08-1602-R-0084 V1.1" the ETR id
-5f18321fb9a41409,KR,KECS-ISIS-0253-2010,has unicode hyphens 8208
+9253ec0d31264194,US,,cert_id missing
 1a8ac8fbe5bfcda9,NO,SERTIT-017,404 for report PDF
 3faae83d248da8ba,IT,,cert_id missing
-d5ed93f6586469f6,CA,503-LSS,regex missing
-a45f04b9a5244a51,KR,KECS-ISIS-0189-2009,has unicode hyphens 8208
-243a29b98f34ae9a,SE,CSEC2019009,space in cert_id
-88eb228eeb487597,KR,KECS-ISIS-0223-2010,has unicode hyphens 8208
-739f48568b127f22,CA,,cert_id missing
-71e471543ea7e03f,CA,383-4-56-CR,regex missing
-ae1b80e9cccec491,JP,Certification No. C0089,not parsed due to PDF issues
-5945cedd8cb01e06,JP,CRP-C0120-01,report and target documents switched
-7e664d089ed905a5,CA,383-4-422P,regex missing
+ae1b80e9cccec491,JP,C0089,pdf issue
+5945cedd8cb01e06,JP,C0120,report and target documents switched
 c508c550607efde5,US,,cert_id missing
-456bfb3665815ccb,KR,KECS-ISIS-0120-2008,has unicode hyphens 8208
-1683817e5db17d9c,JP,Certification No. C0047,not parsed due to PDF issues
-b2aefe8b1bd4d9b3,UK,,cert_id missing
-f0af2383ad08e3ba,SE,CSEC2019011,regex missing
-ae78e8b68bd4c917,US,CCEVS-VR-CC-11074-2020,regex missing
+1683817e5db17d9c,JP,C0047,pdf issue
 323987cac96072cb,TR,14.10.01/10-105,cert_id malformed
 8ac22970bd8042d2,US,,404 for report PDF
-b61b6299f457a01e,SE,CSEC2006001,regex missing
-6fbb34a3a08e943c,CA,,cert_id missing
 6901124edcfe478f,US,CCEVS-VR-03-0038,report and target documents switched
 98668cafb8dda26f,SE,CSEC2019014,report not a PDF (docx)
 d79ecebfe21f2144,US,,404 for report PDF
 7e32023021d5aad2,ES,INF-510,regex missing
 02958808b57334f0,SE,CSEC2019016,report not a PDF (docx)
 391c8b9254ed6ed9,CA,,404 for report PDF
-d3b7b0df0108a466,CA,,cert_id missing
-b17b7e4876e9e482,AU,DXC-EFC-T092-ETR 1.0,
 82fcce2db9ef6063,SG,,404 for report PDF
 5e646214ea5b1044,US,CCEVS-VR-VID11243-2022,has security target uploaded in place of report
-2b2268271ee16bd7,FR,ANSSI-CC 2022/05,not parsed due to PDF issue
-e37e1f81d12b63fd,FR,ANSSI-CC 2022/04,not parsed due to PDF issue
-8b6d2aa804720833,FR,ANSSI-CC-2017/42v2,not parsed due to PDF issue
-614f4dcdee0128bd,IN,IC3S/KOL01/ECITelecom/EAL2/0420/0019/CR,space in cert_id
-3477723044183b31,FR,ANSSI-CC-2017/42v2-S01,regex missing
-a5c2f0adcecf1a33,MY,ISCB-5-RPT-C127-CR-v1.1,regex missing
\ No newline at end of file
+d399d47f958a80dd,AU,Certificate Number: 2020/126,certificate.pdf + OCR
+8b357b0ccdafe13b,AU,Certificate Number: 2019/120,certificate.pdf + OCR
+4fb34c6529549e26,JP,C0064,pdf issue
+c64192c1597a1a14,AU,,cert_id missing
+245f78711cfcb537,JP,C0142,pdf issue
+acd95d9dd4ad410c,AU,Certificate Number: 2019/121,certificate.pdf + OCR
+9903c89cbb35eec1,AU,Certificate Number: 2019/123,certificate.pdf + OCR
+908128ea95f94b09,AU,Certificate Number: 2020/132,certificate.pdf + OCR
+0597940a3fd25600,AU,Certificate Number: 2019/122,certificate.pdf + OCR
+372c24701fa80b5a,AU,Certificate Number: 2019/124,certificate.pdf + OCR
+5db22161b84ab3d5,AU,Certificate Number: 2020/127,certificate.pdf + OCR
+5390c9458174b933,AU,Certificate Number: 2020/129,certificate.pdf + OCR
+bd7d1fa4b1e938e4,AU,Certificate Number: 2020/128,certificate.pdf + OCR
+bd992ce6d76206bf,JP,C0203,pdf issue
+c74dd9d805e44354,FR,ANSSI-CC-2005/12,pdf issue
+ea708fac8f672c0f,JP,C0702,pdf issue
+14e28608791e4a82,AU,,cert_id missing
+96cf0d5d7f98fbf1,AU,Certificate Number: 2020/130,certificate.pdf + OCR
+770f2b13b0723e49,AU,Certificate Number: 2021/134,certificate.pdf + OCR
+3891a6c29ed701b4,FR,ANSSI-CC-2005/02,pdf issue
+7a799fc315ee99da,AU,Certificate Number: 2021/133,certificate.pdf + OCR
+5f6946c674c32b30,AU,Certificate Number: 2022/138,certificate.pdf
+749fae44017016c5,FR,ANSSI-CC-2005/01,pdf issue
+3097b47b6f065ba5,JP,C0062,pdf issue
+9a1c767d358eee50,AU,,cert_id missing
+5ceb859ce72daecf,AU,Certificate Number: 2021/135,certificate.pdf + OCR
+fdf20384e73efb57,AU,Certificate Number: 2020/131,certificate.pdf + OCR
+d2b49eb29a3dcbdc,US,,cert_id missing
+afcf5263d97159f8,AU,Certificate Number: 2019/125,certificate.pdf + OCR
+10d135ab24eab5a8,CA,383-4-270,pdf issue
+4408217c4ce524fd,FR,ANSSI-CC-2003/12,pdf issue
+c3f8e8ddb1dabb02,FR,ANSSI-CC-2003/12,pdf issue
\ No newline at end of file
diff --git a/data/cert_id_eval/random.csv b/data/cert_id_eval/random.csv
new file mode 100644
index 00000000..c995debd
--- /dev/null
+++ b/data/cert_id_eval/random.csv
@@ -0,0 +1,101 @@
+id,cert_id
+61e7578e78fbe755,14.10.06/13-006
+dc91f94aa005a094,CCEVS-VR-VID10789-2017
+0349c893b81f7c1a,NSCIB-CC-0189590-CR
+ece463ca87392f7b,CCEVS-VR-06-0016
+c92a196d0408a5ff,ANSSI-CC-2013/57
+0e22fe4e4e58faf4,BSI-DSZ-CC-1052-V5-2022
+b4bec1e6f04c3533,BSI-DSZ-CC-0478-2008
+940a18f62c79ffae,ANSSI-CC-2022/30
+00fc58baa6297c0d,C0072
+29964f32c68b0ce8,BSI-DSZ-CC-0519-V4-2022
+b2f3874cdb97fe27,BSI-DSZ-CC-0790-2013
+1fb95703e4e12a49,383-4-263
+546f9e2496b1d11c,ANSSI-CC-2015/14
+1dd7f4bf2e677073,ANSSI-CC-2014/20
+b24a14935edd51ad,BSI-DSZ-CC-0783-2013
+302df01b5d463eea,CCEVS-VR-VID10253-2009
+6337d3dcc5362dc6,NSCIB-CC-0122197-CR
+8da378edb071a71c,CCEVS-VR-VID10691-2016
+dd1482a4911a3f1d,CCEVS-VR-05-0105
+99223aca5d9eb3b3,ANSSI-CC-2011/01
+dc45c7a6ef17f8b5,BSI-DSZ-CC-1161-2020
+15fae84c1e6c6c57,ANSSI-CC-2001/05
+eaf0033626287cc4,ANSSI-CC-2012/81
+3357e42cd4dc8605,383-4-256
+4e0dc3133a7e601d,C0513
+8877b69b21e92b41,2019-50-INF-3506
+3c05395ee9e5adf1,BSI-DSZ-CC-0707-2012
+a6e37f57c6e72d8c,ANSSI-CC-2005/13
+f0068ddc5569b772,518-LSS
+04715a4f818f4b71,383-4-345
+3c0185a5cb2d1ea8,CSEC2011001
+f91b60c2954d4b00,CSEC2017005
+b95f2b009d70fad1,BSI-DSZ-CC-0637-2010
+d8c205b4924f91b3,ANSSI-CC-2022/09
+82361ca3849ae025,383-4-420
+40817e88a5a46fd5,CCEVS-VR-VID10541-2014
+c1d88ce9dadd7d2d,BSI-DSZ-CC-0313-2006
+ba27867a74e9c47a,CCEVS-VR-VID10472-2014
+f7cbb33cc639b5b5,KECS-ISIS-0222-2010
+1e7b90184a127017,CCEVS-VR-VID10502-2012,frontpage,
+0d77936727b14c88,BSI-DSZ-CC-0591-2011
+9060684220d85047,ANSSI-CC-2008/01
+b23a595bb30865b5,C0575
+663d89dea15969c8,CRP214
+724d3996cb63d204,CSEC2014001
+5e89afca1d4570a4,C0077
+37c8d23c44b95833,ANSSI-CC-2018/43
+dfb6af00c78cd061,C0421
+b9365d96c9ecdde0,CCEVS-VR-VID10869-2018
+c8427da3733c506d,OCSI/CERT/TEC/05/2017/RC
+780f19df05e0dd18,C0334
+56660124ad6ded6d,BSI-DSZ-CC-1089-2020
+3ba9f330ce93636b,BSI-DSZ-CC-0732-2011
+80932fd09bfc3f98,CCEVS-VR-VID10120-2008
+8298c7814b3b2860,KECS-CISS-1153-2022
+c290ee3692a00006,383-4-438
+642d82f9a5e97fa2,CSEC2017014
+4b2b04554256b523,KECS-ISIS-0117-2008
+4eaceb72085e12f7,C0668
+70b12d34fa092eeb,CCEVS-VR-VID10380-2012
+ec7db499df27dcf9,383-4-397
+61821500cddab34f,BSI-DSZ-CC-0540-2009
+0f3900cdcd0c7f3e,NSCIB-CC-127667-CR3
+238f8edc5eda1358,BSI-DSZ-CC-0285-2005
+b5bcac5eb0f61dfe,CCEVS-VR-10822-2018
+28fb117adbb809e3,CSEC2006002
+c10773ac43ca7c2b,CRP180
+6d6ade44dcc497dd,BSI-DSZ-CC-0312-2005
+c23ca69c2f7528d1,ANSSI-CC-2020/24
+abe89a00b3efa393,C0261
+cf17d0efe1278432,ANSSI-CC-2007/24
+7607e34a7608222c,ISCB-3-RPT-C098-CR-v1
+5ae867bded9ef672,C0180
+3f141b547bd4735f,CCEVS-VR-VID10571-2014
+ed0b1d4c39cbd251,C0727
+abc1f0f55adc2967,CCEVS-VR-VID11049
+d68511e69820f461,CRP260
+59b6bc42781e26d6,CSEC2019006
+d9d0986019774924,KECS-NISS-0050-2006
+318dcee9d8f6ac99,CSEC2018007
+a6fac58198296194,BSI-DSZ-CC-0857-2013
+f1a73b2232787b26,BSI-DSZ-CC-0671-2011
+90f1d708d20deb30,SERTIT-043
+faecd563eb8f1d6c,C0314
+f5d49c63046c69d2,CCEVS-VR-VID10272-2008
+f2e00a1fee27bdd7,BSI-DSZ-CC-0957-2015
+33cb152e0a1be23c,CCEVS-VR-10928-2019
+c16a92b40550193d,383-4-170
+1fb1564dfb0f0b04,ANSSI-CC-2021/26
+8b3c4eb3efd1cdaa,574-LSS
+14ddeb255e0a1cd2,2017-60-INF-2765
+01e3716dd2a5b510,BSI-DSZ-CC-0466-2008
+0c7ef6c32cbdee47,BSI-DSZ-CC-1073-2019
+13e3c6a0cd06ecf6,383-4-340
+814a2ac0aeae08b1,CCEVS-VR-03-0044
+11708925f70ce5d3,Certificate Number: 2011/72
+ca5da2fe138af656,BSI-DSZ-CC-0423-2007
+98937f99289f2499,NSCIB-CC-15-66461-CR
+c3b443f243b95913,CCEVS-VR-11219-2021
+183a5cd3e2aaa0ec,ANSSI-CC-2017/81
\ No newline at end of file
diff --git a/data/cert_id_eval/truth.csv b/data/cert_id_eval/truth.csv
new file mode 100644
index 00000000..80086fc7
--- /dev/null
+++ b/data/cert_id_eval/truth.csv
@@ -0,0 +1,140 @@
+id,cert_id,source,possible
+b2aea954a265af60,559-LSS,frontpage,y
+1380f694ee542194,383-4-343,certificate.docx,n
+b2aefe8b1bd4d9b3,CRP208A,metadata,y
+eaf0033626287cc4,ANSSI-CC-2012/81,frontpage,y
+45aeb28de1e850c3,CCEVS-VR-VID10367-2010,frontpage,y
+5ee9898421c06859,BSI-DSZ-CC-0833-2013,frontpage,y
+f229a8078cb4ff82,BSI-DSZ-CC-1059-V3-2019,frontpage,y
+4b6c4e3f3ee66ff1,ISCB-5-RPT-C126-CR-v1,frontpage,y
+2b6a2b0e304555da,CCEVS-VR-VID10486-2012,frontpage,y
+d72b3643a223681d,SERTIT-074,frontpage,y
+4b9c47a151bcab7b,383-4-179,frontpage,y
+e27eb9efbd67c4f7,,,n
+d8ce75e10487f2f2,C0010,frontpage,y
+4a236a029f71faef,C0236,frontpage,y
+131cd9abf2ecf112,ANSSI-CC-2010/54,frontpage,y
+d2f75cdaeb322e7f,ANSSI-CC-2020/06,frontpage,y
+72427a1b18b32eb7,BSI-DSZ-CC-0839-2013,frontpage,y
+882a9d016c7e1f36,ANSSI-CC-2013/01,frontpage,y
+a0279eb47694a2e8,BSI-DSZ-CC-1116-2020,frontpage,y
+657fe7da80be7a4a,CCEVS-VR-10659-2015,frontpage,y
+f73d08f8cdfe276d,ANSSI-CC-2003/28,frontpage,y
+ad227ad680acd5a0,CSEC2020025,keywords,y
+d5ad4e797ce2a0ca,NSCIB-CC-0059187-CR,frontpage,y
+13c3c43841adf547,CCEVS-VR-02-0027,frontpage,y
+fa384d976e832adb,BSI-DSZ-CC-0811-2013,frontpage,y
+9299f6cd8a266541,BSI-DSZ-CC-0439-2008,frontpage,y
+ae47ae1c2babf63f,383-4-173,frontpage,y
+82ade52e9ef12fd9,ANSSI-CC-2018/32,frontpage,y
+c896595602779acc,CCEVS-VR-VID10402-2013,frontpage,y
+b82b48b0e915aaac,BSI-DSZ-CC-0223-2003,frontpage,y
+5a288eafb7846553,2018-37-INF-2718,frontpage,y
+f6a89befd13a643f,BSI-DSZ-CC-1092-2020,frontpage,y
+92c7355f5ea28e52,CRP243,metadata,y
+0117a4681213926d,C0426,frontpage,y
+c20b0c546ac42cde,C0076,frontpage,y
+c39e6ef4060594e2,BSI-DSZ-CC-0456-2009,frontpage,y
+a42eb7c18fc371fd,CCEVS-VR-10471-2011,frontpage,y
+8ba06ae83d77ef0d,BSI-DSZ-CC-0815-2013,frontpage,y
+3ba9f330ce93636b,BSI-DSZ-CC-0732-2011,frontpage,y
+dc45c7a6ef17f8b5,BSI-DSZ-CC-1161-2020,frontpage,y
+12fd6b1b6c4aecf3,ANSSI-CC-2022/16,frontpage,y
+374b0a920ea4853c,383-4-13,frontpage,y
+7d3546a4000b5b75,ANSSI-CC-2004/08,frontpage,y
+a1f410a18cf45a35,BSI-DSZ-CC-0810-2012,frontpage,y
+e7162eaa3fbb1d31,CCEVS-VR-VID11104-2021,frontpage,y
+ad0230b02113d0fd,ISCB-5-RPT-C010-CR-v1a,frontpage,y
+98a578f7758bb518,BSI-DSZ-CC-0452-2007,frontpage,y
+2c02bf0c17cb16ca,CCEVS-VR-06-0010,frontpage,y
+357b2b1598e71533,CCEVS-VR-VID1092-2017,frontpage,y
+975dae6e02765a90,ANSSI-CC-2013/06,frontpage,y
+74e6430d8c4ea5b3,KECS-NISS-0068-2007,frontpage,y
+db487c4723ff5829,383-4-328,frontpage,y
+9646e10d3f2b5cd1,CCEVS-VR-10919-2018,frontpage,y
+5498bfcecf19dfa3,ANSSI-CC-2021/37,frontpage,y
+2b665984895290b0,CSEC2018001,keywords,y
+611a1cc1a2d405d3,BSI-DSZ-CC-0822-V3-2021,frontpage,y
+40640508ac3a37f2,21.0.03/TSE-CCCS-50,frontpage,y
+639a4782da0e8ecf,CCEVS-VR-10635-2015,frontpage,y
+420876ec1e5ba657,ANSSI-CC-2012/71,frontpage,y
+bdf8d5624e230986,BSI-DSZ-CC-0809-V2-2016,frontpage,y
+cf1c544add4bf860,BSI-DSZ-CC-1137-V2-2021,frontpage,y
+5360e7547186859e,383-4-475,frontpage,y
+a69ec271f8651d1b,BSI-DSZ-CC-1148-2020,frontpage,y
+ffe4328dcdc6b538,BSI-DSZ-CC-0793-2012,frontpage,y
+83a723fec25495e7,CSEC2017011,keywords,y
+16513a1bff79b46a,BSI-DSZ-CC-1044-V3-2020,frontpage,y
+a93edea77d9f3338,BSI-DSZ-CC-0582-2009,frontpage,y
+f0af2383ad08e3ba,CSEC2019011,frontpage,y
+031667f4e242da61,BSI-DSZ-CC-0588-2009,frontpage,y
+90d4421606e3b32f,KECS-CISS-0230-2010,frontpage,y
+8f85d66508aa4148,CCEVS-VR-VID11107-2021,frontpage,y
+cc78a842a1cc2c8b,CRP238,frontpage,y
+7c7e48e3aed8bfd4,ANSSI-CC-2020/60,frontpage,y
+2bfb05dec3fb0bb3,CCEVS-VR-07-0009,frontpage,y
+422fb9723d7b0397,CCEVS-VR-VID10754-2017,frontpage,y
+e37540482473cc09,NSCIB-CC-0490158-CR,frontpage,y
+0dcc691ea04d1b52,383-4-351,certificate.docx,n
+e56d5d27cb2b3e6f,BSI-DSZ-CC-0579-2009,frontpage,y
+4fa8af0edccb7ada,BSI-DSZ-CC-0218-2003,frontpage,y
+eb68ee4c8fc8b966,C0220,frontpage,y
+bff1330c13ccd99b,BSI-DSZ-CC-0347-2006,frontpage,y
+9c6440ecec046cc0,Certification Report 2012/79,frontpage,y
+205dba83d0a6e46c,2011-8-INF-835,frontpage,y
+b9f476ad0695d397,C0514,frontpage,y
+b0d820cc80fa507b,565-LSS,frontpage,y
+f5e2a07ae4043fca,C0470,frontpage,y
+0431530b3c3d1973,Certificate Number: 2009/61,frontpage,y
+afcf5263d97159f8,Certificate Number: 2019/125,certificate.pdf,y
+562ae7baded055ab,383-4-296,frontpage,y
+921ecc7c3e5f4884,BSI-DSZ-CC-0477-2007,frontpage,y
+d3420c2bf8974a5d,BSI-DSZ-CC-0968-2016,frontpage,y
+c24c5157e8af96d4,ANSSI-CC-2005/21,frontpage,y
+406b72e27c0d6c5d,383-4-175,frontpage,y
+cde736a2a31b7298,CSEC2016012,keywords,y
+b131130935031b4d,Certificate Number: 2002/22,frontpage,y
+22066b781cc7d1b0,C0568,frontpage,y
+dbe20d7c305b24eb,NSCIB-CC-13-37760-CR2,frontpage,y
+6c14f733fe7aa28e,NSCIB-CC-66030-CR5,frontpage,y
+74a8ddae1bb0e894,383-4-10,frontpage,y
+b96dff73aed97fdf,2019-48-INF-3811,frontpage,y
+20c98a49ea86a519,CRP153,frontpage,y
+10dc636e9da3e1b6,383-4-349,filename,y
+e2ecf0daf73716f2,511-LSS,frontpage,y
+9219f2fb5f305269,KECS-ISIS-0119-2008,frontpage,y
+10339846e519597c,516-LSS,frontpage,y
+f0068ddc5569b772,518-LSS,frontpage,y
+2d05e976e1adf696,383-4-411,filename,y
+28fb117adbb809e3,CSEC2006002,frontpage,y
+456bfb3665815ccb,KECS-ISIS-0120-2008,frontpage,y
+614f4dcdee0128bd,IC3S/KOL01/ECITelecom/EAL2/0420/0019/CR,frontpage,y
+a79f15874e255007,CRP153,frontpage,y
+bfc86f14e2484335,383-4-430,filename,y
+32d4bf275131aa15,KECS-ISIS-0118-2008,frontpage,y
+1491021aa321b828,517-CCS-2020,frontpage,y
+a91b4c64329d05a8,CSEC2019010,frontpage,y
+c5bad5b309469e22,519-EWA,frontpage,y
+876cb2fe02017f3f,C0012,frontpage,y
+f9c8da9deff77ab5,383-4-384,filename,y
+6266d99d3b983091,383-4-347,filename,y
+4de4a9f436958574,525-LSS-2020,filename,y
+7e664d089ed905a5,383-4-422,frontpage,y
+ae78e8b68bd4c917,CCEVS-VR-CC-11074-2020,frontpage,y
+b61b6299f457a01e,CSEC2006001,frontpage,y
+a5c2f0adcecf1a33,ISCB-5-RPT-C127-CR-v1.1,frontpage,y
+3477723044183b31,ANSSI-CC-2017/42v2-S01,frontpage,y
+d3b7b0df0108a466,383-4-344,filename,y
+6fbb34a3a08e943c,383-4-334,filename,y
+b2aefe8b1bd4d9b3,CRP208A,filename,y
+c312bbf9e303b844,CSEC2019012,frontpage,y
+fb6d7faf0ed2dca5,383-4-362,filename,y
+b3ae94eeccc566ba,383-4-17,frontpage,y
+5f18321fb9a41409,KECS-ISIS-0253-2010,frontpage,y
+a45f04b9a5244a51,KECS-ISIS-0189-2009,frontpage,y
+88eb228eeb487597,KECS-ISIS-0223-2010,frontpage,y
+d5ed93f6586469f6,503-LSS,frontpage,y
+243a29b98f34ae9a,CSEC2019009,frontpage,y
+739f48568b127f22,383-4-346,filename,y
+71e471543ea7e03f,383-4-56,frontpage,y
+f0af2383ad08e3ba,CSEC2019011,frontpage,y
\ No newline at end of file
diff --git a/notebooks/cc/references.ipynb b/notebooks/cc/references.ipynb
index a55628c4..bafb7edf 100644
--- a/notebooks/cc/references.ipynb
+++ b/notebooks/cc/references.ipynb
@@ -698,4 +698,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 1
-}
+}
\ No newline at end of file
diff --git a/notebooks/cert_id_eval.ipynb b/notebooks/cert_id_eval.ipynb
index c0463750..c9ed32a8 100644
--- a/notebooks/cert_id_eval.ipynb
+++ b/notebooks/cert_id_eval.ipynb
@@ -18,7 +18,8 @@
     "1. Certificates with no id assigned.\n",
     "2. Duplicate certificate id assignments (when two certificates get the same ID assigned).\n",
     "3. Certificates that have the same certification report document (an issue of the input data that we get\n",
-    "   that explains some of the duplicate certificate id assignments)."
+    "   that explains some of the duplicate certificate id assignments).\n",
+    "4. Compares a random sample of certificates with assigned ground truth ID."
    ]
   },
   {
@@ -32,7 +33,8 @@
    "outputs": [],
    "source": [
     "from sec_certs.dataset import CCDataset\n",
-    "import csv"
+    "import csv\n",
+    "import pandas as pd"
    ]
   },
   {
@@ -41,14 +43,18 @@
    "metadata": {
     "pycharm": {
      "name": "#%%\n"
-    }
+    },
+    "scrolled": true
    },
    "outputs": [],
    "source": [
     "# dset = CCDataset.from_web_latest()\n",
     "dset = CCDataset.from_json(\"../cc_dset/CommonCriteria_dataset.json\")\n",
-    "# dset._compute_normalized_cert_ids()\n",
-    "# dset.to_json()"
+    "#dset._download_reports(False)\n",
+    "#dset._convert_reports_to_txt(False)\n",
+    "#dset._extract_report_keywords()\n",
+    "#dset._compute_normalized_cert_ids()\n",
+    "#dset.to_json()"
    ]
   },
   {
@@ -59,7 +65,7 @@
     }
    },
    "source": [
-    "## Certificates with no id\n",
+    "## 1. Certificates with no id\n",
     "\n",
     "Here we report the number of certificates in our dataset that we have no certificate ID for."
    ]
@@ -76,11 +82,12 @@
    "outputs": [],
    "source": [
     "missing_id_dgsts = set()\n",
+    "missing_id = []\n",
     "for cert in dset:\n",
     "    if not cert.heuristics.cert_id:\n",
-    "        print(cert.dgst, cert.heuristics.cert_id, cert.scheme)\n",
     "        missing_id_dgsts.add(cert.dgst)\n",
-    "print(f\"Total: {len(missing_id_dgsts)}\")"
+    "        missing_id.append((cert.dgst, cert.scheme))\n",
+    "pd.DataFrame(missing_id, columns=[\"id\", \"scheme\"])"
    ]
   },
   {
@@ -94,36 +101,12 @@
    "cell_type": "code",
    "execution_count": null,
    "metadata": {
-    "pycharm": {
-     "name": "#%%\n"
-    }
+    "scrolled": false
    },
    "outputs": [],
    "source": [
-    "i = 0\n",
-    "with open(\"../data/cert_id_eval/missing_ids.csv\", \"r\") as f:\n",
-    "    reader = csv.DictReader(f)\n",
-    "    for line in reader:\n",
-    "        try:\n",
-    "            cert = dset[line[\"id\"]]\n",
-    "        except:\n",
-    "            continue\n",
-    "        if line[\"cert_id\"] and line[\"cert_id\"] != cert.heuristics.cert_id:\n",
-    "            i += 1\n",
-    "            print(line[\"id\"], line[\"cert_id\"], cert.heuristics.cert_id, line[\"reason\"])\n",
-    "print(f\"Total: {i}\")"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {
-    "pycharm": {
-     "name": "#%% md\n"
-    }
-   },
-   "source": [
-    "\n",
-    "The following cell checks which manually analyzed missing certificate IDs were since fixed."
+    "missing_manual = pd.read_csv(\"../data/cert_id_eval/missing_ids.csv\")\n",
+    "set(missing_manual.id) == missing_id_dgsts"
    ]
   },
   {
@@ -136,47 +119,37 @@
    },
    "outputs": [],
    "source": [
-    "manual_missing_ids = set()\n",
-    "i = 0\n",
-    "with open(\"../data/cert_id_eval/missing_ids.csv\", \"r\") as f:\n",
-    "    reader = csv.DictReader(f)\n",
-    "    for line in reader:\n",
-    "        manual_missing_ids.add(line[\"id\"])\n",
-    "        if line[\"id\"] not in missing_id_dgsts:\n",
-    "            i += 1\n",
-    "            print(\",\".join(line.values()))\n",
-    "print(f\"Total: {i}\")"
+    "num_missing_manual = missing_manual.shape[0]\n",
+    "num_missing_manual_fixable = missing_manual.cert_id.count()\n",
+    "num_missing_manual_unfixable = num_missing_manual - num_missing_manual_fixable\n",
+    "print(f\"\\\\newcommand{{\\\\numccmissingid}}{{{num_missing_manual}}}\")\n",
+    "print(f\"\\\\newcommand{{\\\\numccmissingidfixable}}{{{num_missing_manual_fixable}}}\")\n",
+    "print(f\"\\\\newcommand{{\\\\numccmissingidunfixable}}{{{num_missing_manual_unfixable}}}\")"
    ]
   },
   {
-   "cell_type": "markdown",
+   "cell_type": "code",
+   "execution_count": null,
    "metadata": {},
+   "outputs": [],
    "source": [
-    "The following cell lists missing certificate IDs that *went missing* since manual analysis."
+    "missing_manual.loc[missing_manual.cert_id.isnull()].reason.value_counts()"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": null,
-   "metadata": {
-    "pycharm": {
-     "name": "#%%\n"
-    }
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
-    "new_missing_ids = missing_id_dgsts.difference(manual_missing_ids)\n",
-    "for idd in new_missing_ids:\n",
-    "    cert = dset[idd]\n",
-    "    print(idd, cert.heuristics.cert_id, cert.pdf_data.report_filename)\n",
-    "print(f\"Total: {len(new_missing_ids)}\")"
+    "missing_manual.loc[missing_manual.cert_id.notnull()].reason.value_counts()"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Duplicate certificate id assignment\n",
+    "## 2. Duplicate certificate id assignment\n",
     "\n",
     "Here we report the number of certificates in our dataset that have a duplicate certiticate\n",
     "ID assigned."
@@ -202,8 +175,7 @@
     "for idd, entries in id_mapping.items():\n",
     "    if len(entries) > 1 and idd:\n",
     "        print(idd, entries)\n",
-    "        duplicate_id_dgsts.update(entries)\n",
-    "print(f\"Total: {len(duplicate_id_dgsts)}\")"
+    "        duplicate_id_dgsts.update(entries)"
    ]
   },
   {
@@ -214,7 +186,7 @@
     }
    },
    "source": [
-    "## Duplicate report documents\n",
+    "## 3. Duplicate report documents\n",
     "\n",
     "Some certificates have erroneously uploaded certificate reports, here we check their\n",
     "hashes and report such duplicates in the input data."
@@ -243,7 +215,9 @@
     "        print(hash, entries)\n",
     "        for entry in entries:\n",
     "            duplicate_doc_dgsts.add(entry)\n",
-    "print(f\"Total: {len(duplicate_doc_dgsts)}\")"
+    "            \n",
+    "duplicate_ids_due_doc = duplicate_doc_dgsts.intersection(duplicate_id_dgsts)\n",
+    "duplicate_ids_issue = duplicate_id_dgsts.difference(duplicate_doc_dgsts)"
    ]
   },
   {
@@ -254,8 +228,7 @@
     }
    },
    "source": [
-    "The following prints the amount of certificate id duplicates due to input data (two or more\n",
-    "certificates share a certification report document)."
+    "The following prints the amount of certificate id duplicates that are not due to input data (and are really our problem)."
    ]
   },
   {
@@ -268,8 +241,19 @@
    },
    "outputs": [],
    "source": [
-    "duplicate_ids_due_doc = duplicate_doc_dgsts.intersection(duplicate_id_dgsts)\n",
-    "print(len(duplicate_ids_due_doc))"
+    "for id in duplicate_ids_issue:\n",
+    "    print(id, dset[id].heuristics.cert_id)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "print(f\"\\\\newcommand{{\\\\numccduplicateid}}{{{len(duplicate_id_dgsts)}}}\")\n",
+    "print(f\"\\\\newcommand{{\\\\numccduplicateidcolission}}{{{len(duplicate_ids_due_doc)}}}\")\n",
+    "print(f\"\\\\newcommand{{\\\\numccduplicateidissue}}{{{len(duplicate_ids_issue)}}}\")"
    ]
   },
   {
@@ -280,7 +264,7 @@
     }
    },
    "source": [
-    "The following prints the amount of certificate id duplicates that are not due to input data."
+    "### Check manually evaluated duplicates"
    ]
   },
   {
@@ -289,50 +273,42 @@
    "metadata": {
     "pycharm": {
      "name": "#%%\n"
-    }
+    },
+    "scrolled": false
    },
    "outputs": [],
    "source": [
-    "duplicate_ids_issue = duplicate_id_dgsts.difference(duplicate_doc_dgsts)\n",
-    "print(len(duplicate_ids_issue))"
+    "duplicate_manual = pd.read_csv(\"../data/cert_id_eval/duplicate_ids_new.csv\")\n",
+    "set(duplicate_manual.id) == duplicate_id_dgsts"
    ]
   },
   {
-   "cell_type": "markdown",
-   "metadata": {
-    "pycharm": {
-     "name": "#%% md\n"
-    }
-   },
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
    "source": [
-    "### Check manually evaluated duplicates\n",
-    "\n",
-    "The following cell checks that id collisions that were manually analyzed in the past have been fixed.\n",
-    "A `True` means that we have now assigned the correct ID."
+    "duplicate_manual[duplicate_manual.result == \"tp\"].reason.value_counts()"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "duplicate_manual[duplicate_manual.result == \"fp\"].reason.value_counts()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
    "metadata": {
     "pycharm": {
-     "name": "#%%\n"
+     "name": "#%% md\n"
     }
    },
-   "outputs": [],
    "source": [
-    "i = 0\n",
-    "with open(\"../data/cert_id_eval/duplicate_ids.csv\", \"r\") as f:\n",
-    "    reader = csv.DictReader(f)\n",
-    "    for line in reader:\n",
-    "        try:\n",
-    "            cert = dset[line[\"id\"]]\n",
-    "        except:\n",
-    "            continue\n",
-    "        if line[\"true_id\"] != cert.heuristics.cert_id:\n",
-    "            print(line[\"id\"],line[\"result\"], line[\"true_id\"] == cert.heuristics.cert_id , line[\"true_id\"], cert.heuristics.cert_id, line[\"fixable\"])\n",
-    "            i += 1\n",
-    "print(f\"Total: {i}\")"
+    "The following cell lists those duplicates that were fixed by changes since manual analysis."
    ]
   },
   {
@@ -343,7 +319,7 @@
     }
    },
    "source": [
-    "The following cell lists those duplicates that were fixed by changes since manual analysis."
+    "## 4. Manually assigned ground truth comparison"
    ]
   },
   {
@@ -356,27 +332,46 @@
    },
    "outputs": [],
    "source": [
-    "manual_duplicate_ids = set()\n",
-    "i = 0\n",
-    "with open(\"../data/cert_id_eval/duplicate_ids.csv\", \"r\") as f:\n",
+    "correct = set()\n",
+    "possible = set()\n",
+    "impossible = set()\n",
+    "with open(\"../data/cert_id_eval/truth.csv\", \"r\") as f:\n",
     "    reader = csv.DictReader(f)\n",
     "    for line in reader:\n",
-    "        manual_duplicate_ids.add(line[\"id\"])\n",
-    "        if line[\"id\"] not in duplicate_id_dgsts:\n",
-    "            i += 1\n",
-    "            print(\",\".join(line.values()))\n",
-    "print(f\"Total: {i}\")"
+    "        cert = dset[line[\"id\"]] # and (line[\"cert_id\"] or cert.heuristics.cert_id is not None)\n",
+    "        if line[\"cert_id\"] != cert.heuristics.cert_id:\n",
+    "            print(line[\"id\"], line[\"cert_id\"], cert.heuristics.cert_id, line[\"source\"], line[\"possible\"])\n",
+    "            if line[\"possible\"] == \"y\":\n",
+    "                possible.add(line[\"id\"])\n",
+    "            else:\n",
+    "                impossible.add(line[\"id\"])\n",
+    "        else:\n",
+    "            correct.add(line[\"id\"])\n",
+    "print(len(correct), len(possible), len(impossible))"
    ]
   },
   {
-   "cell_type": "markdown",
+   "cell_type": "code",
+   "execution_count": null,
    "metadata": {
     "pycharm": {
-     "name": "#%% md\n"
+     "name": "#%%\n"
     }
    },
+   "outputs": [],
    "source": [
-    "The following cell lists duplicates that were *created* since manual analysis."
+    "correct = set()\n",
+    "incorrect = set()\n",
+    "with open(\"../data/cert_id_eval/random.csv\", \"r\") as f:\n",
+    "    reader = csv.DictReader(f)\n",
+    "    for line in reader:\n",
+    "        cert = dset[line[\"id\"]]\n",
+    "        if line[\"cert_id\"] != cert.heuristics.cert_id:\n",
+    "            print(line[\"id\"], line[\"cert_id\"], cert.heuristics.cert_id)\n",
+    "            incorrect.add(line[\"id\"])\n",
+    "        else:\n",
+    "            correct.add(line[\"id\"])\n",
+    "print(len(correct), len(incorrect))"
    ]
   },
   {
@@ -389,11 +384,9 @@
    },
    "outputs": [],
    "source": [
-    "new_duplicate_ids = duplicate_id_dgsts.difference(manual_duplicate_ids)\n",
-    "for idd in new_duplicate_ids:\n",
-    "    cert = dset[idd]\n",
-    "    print(idd, cert.heuristics.cert_id)\n",
-    "print(f\"Total: {len(new_duplicate_ids)}\")"
+    "print(f\"\\\\newcommand{{\\\\numccideval}}{{{len(correct) + len(incorrect)}}}\")\n",
+    "print(f\"\\\\newcommand{{\\\\numccidevalcorrect}}{{{len(correct)}}}\")\n",
+    "print(f\"\\\\newcommand{{\\\\numccidevalincorrect}}{{{len(incorrect)}}}\")"
    ]
   },
   {
@@ -425,4 +418,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 1
-}
\ No newline at end of file
+}
diff --git a/sec_certs/rules.yaml b/sec_certs/rules.yaml
index 6c481629..d0204fd5 100644
--- a/sec_certs/rules.yaml
+++ b/sec_certs/rules.yaml
@@ -12,11 +12,11 @@ cc_cert_id:
     - "BSI-DSZ-CC-[0-9]+-[0-9][0-9][0-9][0-9]"  # German BSI (number + year, no version)
     - "BSI-DSZ-CC-[0-9]+-(?:V|v)[0-9]+(?!-)"  # German BSI (number + version but no year => no - after version)
   FR:
-    - "ANSSI(?:-|-CC-|-CC )[0-9]+?/[0-9]+(v[1-9])?"  # French
-    - "ANSS[Ii]-CC-[0-9][0-9][0-9][0-9][/-_][0-9][0-9]+(?!-M|-S|-R)"  # French (/two or more digits then NOT -M or -S)
-    - "ANSS[Ii]-CC-[0-9][0-9][0-9][0-9][/-_][0-9]+(?:v[0-9])?[_/-][MSR][0-9]+"  # French, maintenance or surveillance report (ANSSI-CC-2014_46_M01)
+    - "ANSS[Ii](?:-|-CC-|-CC )[0-9]{4}/[0-9]+(v[1-9])?"  # French
+    - "ANSS[Ii]-CC[ -][0-9]{4}[/-_][0-9][0-9]+(?!-M|-S|-R)"  # French (/two or more digits then NOT -M or -S)
+    - "ANSS[Ii]-CC[ -][0-9]{4}[/-_][0-9]+(?:v[0-9])?[_/-][MSR][0-9]+"  # French, maintenance or surveillance report (ANSSI-CC-2014_46_M01)
     # 'ANSSI-CC-CER-F-.+?', # French
-    - "DCSSI-[0-9]+/[0-9]+"  # French (DCSSI-2009/07)
+    - "DCSS[Ii]-[0-9]+/[0-9]+"  # French (DCSSI-2009/07)
     - "Certification Report [0-9]+/[0-9]+" # French or Australia! Solved because we limit ourselves to scheme when doing heuristics.
     - "Rapport de certification [0-9]+/[0-9]+"  # French
   NL:
@@ -38,8 +38,8 @@ cc_cert_id:
     - "[0-9][0-9][0-9][ -](?:EWA|LSS|CCS)(?:[ -]20[0-9][0-9])?"  # Canada (522-EWA-2020, 524 LSS 2020, 503-LSS)
     - "[0-9][0-9][0-9](?:%20|-)(?:EWA|LSS|CCS)(?:%20|-)(?:20[0-9][0-9]%20|)CR%20v[0-9]\\.[0-9]" # Canada filename with space (518-LSS%20CR%20v1.0)
   UK:
-    - "CRP[0-9]+"  # UK CESG
-    - "CERTIFICATION REPORT No. P[0-9]+"  # UK CESG
+    - "CRP[0-9]+[A-Z]?"  # UK CESG
+    - "CERTIFICATION REPORT No. P[0-9]+[A-Z]?"  # UK CESG
   ES:
     - "20[0-9][0-9][-‐][0-9]+[-‐]INF[-‐][0-9]+([-‐]?[ -‐](?:V|v)[0-9]+)?"  # Spain ("2006-4-INF-98 v2" or "2006-4-INF-98-v2" or "2020-34-INF-3784- v1")
   KR:
diff --git a/sec_certs/sample/cc_certificate_id.py b/sec_certs/sample/cc_certificate_id.py
index 3034d294..28016ee5 100644
--- a/sec_certs/sample/cc_certificate_id.py
+++ b/sec_certs/sample/cc_certificate_id.py
@@ -1,3 +1,4 @@
+import re
 from dataclasses import dataclass
 from typing import List, Tuple
 
@@ -12,27 +13,15 @@ class CertificateId:
     raw: str
 
     def _canonical_fr(self) -> str:
-        # TODO: Unify ANSSI vs DCSSI vs Certification Report vs Rapport de certification
         new_cert_id = self.clean
-
-        # This currently only handles the ANSSI-CC-0000... ids.
-        if not new_cert_id.startswith("ANSS"):
-            return new_cert_id
-
-        if new_cert_id.startswith("ANSSi"):  # mistyped ANSSi
-            new_cert_id = "ANSSI" + new_cert_id[4:]
-
-        # Bug - getting out of index - ANSSI-2009/30
-        # TMP solution
-        # TODO: Fix me, @georgefi
-        if len(new_cert_id) >= len("ANSSI-CC-0000") + 1:
-            if (
-                new_cert_id[len("ANSSI-CC-0000")] == "_"
-            ):  # _ instead of / after year (ANSSI-CC-2010_40 -> ANSSI-CC-2010/40)
-                new_cert_id = new_cert_id[: len("ANSSI-CC-0000")] + "/" + new_cert_id[len("ANSSI-CC-0000") + 1 :]
-
-        if "_" in new_cert_id:  # _ instead of -
-            new_cert_id = new_cert_id.replace("_", "-")
+        rules = [
+            "(?:Rapport de certification|Certification Report) ([0-9]+[/-_][0-9]+(?:v[1-9])?(?:[_/-][MSR][0-9]+)?)",
+            "(?:ANSS[Ii]|DCSSI)(?:-CC)?[- ]([0-9]+[/-_][0-9]+(?:v[1-9])?(?:[_/-][MSR][0-9]+)?)",
+            "([0-9]+[/-_][0-9]+(?:v[1-9])?(?:[_/-][MSR][0-9]+)?)",
+        ]
+        for rule in rules:
+            if match := re.match(rule, new_cert_id):
+                return "ANSSI-CC-" + match.group(1).replace("_", "/")
 
         return new_cert_id
 
@@ -54,19 +43,9 @@ class CertificateId:
 
             return cert_num, cert_version, cert_year
 
-        # start_year = 1996
-        # limit_year = datetime.now().year + 1
         bsi_parts = self.clean.split("-")
 
         cert_num, cert_version, cert_year = extract_parts(bsi_parts)
-        # if cert_year is None:
-        #     for year in range(start_year, limit_year):
-        #         cert_id_possible = cert_id + "-" + str(year)
-        #
-        #         if cert_id_possible in all_cert_ids:
-        #             # we found version with year
-        #             cert_year = str(year)
-        #             break
 
         # reconstruct BSI number again
         new_cert_id = "BSI-DSZ-CC"
@@ -83,15 +62,15 @@ class CertificateId:
         cert_id = self.clean
         spain_parts = cert_id.split("-")
         cert_year = spain_parts[0]
-        cert_batch = spain_parts[1]
-        cert_num = spain_parts[3]
+        cert_batch = spain_parts[1].lstrip("0")
+        cert_num = spain_parts[3].lstrip("0")
 
         if "v" in cert_num:
             cert_num = cert_num[: cert_num.find("v")]
         if "V" in cert_num:
             cert_num = cert_num[: cert_num.find("V")]
 
-        new_cert_id = f"{cert_year}-{cert_batch}-INF-{cert_num.strip()}"  # drop version
+        new_cert_id = f"{cert_year}-{cert_batch}-INF-{cert_num.strip()}"  # drop version # TODO: Maybe do not drop?
 
         return new_cert_id
 
@@ -108,6 +87,27 @@ class CertificateId:
     def _canonical_se(self):
         return self.clean.replace(" ", "")
 
+    def _canonical_uk(self):
+        new_cert_id = self.clean
+        if match := re.match("CERTIFICATION REPORT No. P([0-9]+[A-Z]?)", new_cert_id):
+            new_cert_id = "CRP" + match.group(1)
+        return new_cert_id
+
+    def _canonical_ca(self):
+        new_cert_id = self.clean
+        if new_cert_id.endswith("-CR"):
+            new_cert_id = new_cert_id[:-3]
+        if new_cert_id.endswith("P"):
+            new_cert_id = new_cert_id[:-1]
+        return new_cert_id.replace(" ", "-")
+
+    def _canonical_jp(self):
+        new_cert_id = self.clean
+        if match := re.match("Certification No. (C[0-9]+)", new_cert_id):
+            return match.group(1)
+        if match := re.search("CRP-(C[0-9]+)-", new_cert_id):
+            return match.group(1)
+
     @property
     def clean(self) -> str:
         """
@@ -128,12 +128,16 @@ class CertificateId:
             "IT": self._canonical_it,
             "IN": self._canonical_in,
             "SE": self._canonical_se,
-            # TODO: Unify UK CRP... vs Certification REPORT No.
-            # TODO: Unify JP C0000 vs JISEC-...
-            # TODO: Unify US (-CR and no -CR)
+            "UK": self._canonical_uk,
+            "CA": self._canonical_ca,
+            "JP": self._canonical_jp,
         }
 
         if self.scheme in schemes:
             return schemes[self.scheme]()
         else:
             return self.clean
+
+
+def canonicalize(cert_id_str: str, scheme: str) -> str:
+    return CertificateId(scheme, cert_id_str).canonical
diff --git a/sec_certs/sample/common_criteria.py b/sec_certs/sample/common_criteria.py
index 53afd9e2..a950ee53 100644
--- a/sec_certs/sample/common_criteria.py
+++ b/sec_certs/sample/common_criteria.py
@@ -25,7 +25,7 @@ from sec_certs.cert_rules import (
     rules,
     security_level_csv_scan,
 )
-from sec_certs.sample.cc_certificate_id import CertificateId
+from sec_certs.sample.cc_certificate_id import canonicalize
 from sec_certs.sample.certificate import Certificate
 from sec_certs.sample.certificate import Heuristics as BaseHeuristics
 from sec_certs.sample.certificate import References, logger
@@ -308,7 +308,7 @@ class CommonCriteriaCert(
                     matches[cert_id] += 1
             if not matches:
                 return {}
-            total = sum(matches.values())
+            total = max(matches.values())
             results = {}
             for candidate, count in matches.items():
                 results[candidate] = count / total
@@ -330,7 +330,7 @@ class CommonCriteriaCert(
             matches: Counter = Counter(cert_id_matches[scheme])
             if not matches:
                 return {}
-            total = sum(matches.values())
+            total = max(matches.values())
             results = {}
             for candidate, count in matches.items():
                 results[candidate] = count / total
@@ -355,7 +355,7 @@ class CommonCriteriaCert(
                         matches[cert_id] += 1
             if not matches:
                 return {}
-            total = sum(matches.values())
+            total = max(matches.values())
             results = {}
             for candidate, count in matches.items():
                 results[candidate] = count / total
@@ -368,16 +368,18 @@ class CommonCriteriaCert(
             filename_id = self.filename_cert_id(scheme)
             keywords_id = self.keywords_cert_id(scheme)
 
-            # Join them and weigh them, each is normalized to sum weights to 1 (if anything is returned)
+            # Join them and weigh them, each is normalized with weights from 0 to 1 (if anything is returned)
             candidates: Dict[str, float] = defaultdict(lambda: 0.0)
+            # TODO: Add heuristic based on ordering of ids (and extracted year + increment)
+            # TODO: Add heuristic based on length
             for candidate, count in frontpage_id.items():
-                candidates[candidate] += count * 1.5
+                candidates[canonicalize(candidate, scheme)] += count * 1.5
             for candidate, count in metadata_id.items():
-                candidates[candidate] += count * 1.2
-            for candidate, count in filename_id.items():
-                candidates[candidate] += count * 1.2
+                candidates[canonicalize(candidate, scheme)] += count * 1.2
             for candidate, count in keywords_id.items():
-                candidates[candidate] += count * 1.0
+                candidates[canonicalize(candidate, scheme)] += count * 1.0
+            for candidate, count in filename_id.items():
+                candidates[canonicalize(candidate, scheme)] += count * 1.0
             return candidates
 
     @dataclass
@@ -987,7 +989,7 @@ class CommonCriteriaCert(
         candidates = self.pdf_data.candidate_cert_ids(self.scheme)
 
         if candidates:
-            candidate = max(candidates, key=candidates.get)
-            # And finally make it canonical, but only if we have it
-            cert_id = CertificateId(self.scheme, candidate)
-            self.heuristics.cert_id = cert_id.canonical
+            max_weight = max(candidates.values())
+            max_candidates = list(filter(lambda x: candidates[x] == max_weight, candidates.keys()))
+            max_candidates.sort(key=len, reverse=True)
+            self.heuristics.cert_id = max_candidates[0]