Datasets:
prompt stringlengths 60 2.89k | chosen stringlengths 151 5.74k | rejected stringlengths 14 4.5k | source stringclasses 52
values | defect stringclasses 12
values | negatives stringclasses 3
values |
|---|---|---|---|---|---|
A crowdfunding platform is paying out to 'Ahmed Moussa' (Egyptian, national ID 28501011234567). The UN list has 'Ahmed H. Moussa' with no DOB but a similar ID. The parent's name is the same. What constitutes a match? | The partial ID match ('similar ID') is a strong indicator—Egyptian national IDs encode DOB and governorate, so a similar ID suggests the same person. The parent's name ('H.' as middle name) matching is corroborative, as the UN list likely uses a patronymic. The absence of DOB is not an issue because the ID itself is a ... | The parent's name ('H.' as middle name) matching is a strong indicator, as the UN list likely uses a patronymic. The absence of DOB is not an issue because the parent's name is a reliable familial link. The partial ID match ('similar ID') is corroborative but secondary, since Egyptian national IDs encode DOB and govern... | synthetic:deepseek/deepseek-v4-flash-0731 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
Our vendor name is 'Redline Logistics Ltd' but the list has 'Red Line Logistics' without the 'Ltd'. Is a missing corporate suffix and a space difference enough to match, or do we need an address? | Yes, a missing corporate suffix and a space difference are enough to match, and we do not need an address to make that determination. 'Ltd' is a legal form indicator that is often omitted in list entries; 'Redline' vs 'Red Line' is a common compound-word variation (one word vs two). These are minor spelling and formatt... | Yes, a missing corporate suffix and a space difference are enough to match, and we do not need an address to make that determination. 'Ltd' is a legal form indicator that is often omitted in list entries; 'Redline' vs 'Red Line' is a common compound-word variation (one word vs two). These are minor spelling and formatt... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
Individual: 'Ivan Ivanov' with passport number 123456; OFAC list has 'Ivan Ivanov' but with a different passport number (789012) and same DOB. Passport mismatch but name and DOB match. Is this a true match or a false positive? Justify. | This is a true match. The name 'Ivan Ivanov' is exact, and DOB matches—these are highly corroborative. Passport numbers can be reissued, mistyped, or differ due to passport renewal; a mismatch alone does not disprove identity. The probability that two different people share the same full name and DOB is low, especially... | This is a true match. The name 'Ivan Ivanov' is exact, and DOB matches—these are highly corroborative. Passport numbers can be reissued, mistyped, or differ due to passport renewal; a mismatch alone does not disprove identity. The probability that two different people share the same full name and DOB is low, especially... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
A luxury car dealership in Monte Carlo is checking a buyer, 'Li Wei' (Chinese, passport E55667788, DOB 1975-05-05). The UN list entry for 'Li Wei' has no DOB but includes a passport number that is a one-digit transposition of this one. What is the match decision? | The name 'Li Wei' is an exact match, and the passport number is a transposition—a common data-entry error. Since the UN list entry lacks a DOB, the passport number is the only corroborating identifier, and a one-digit transposition is a very weak distance (edit distance 1) that is more likely a typo than a different pe... | The name 'Li Wei' is an exact match, and the passport number is a transposition—a common data-entry error. Since the UN list entry lacks a DOB, the passport number is the only corroborating identifier, and a one-digit transposition is a very weak distance (edit distance 1) that is more likely a typo than a different pe... | synthetic:deepseek/deepseek-v4-flash-0731 | factual_error | defect-injection:deepseek/deepseek-v4-flash-0731 |
While vetting a cargo manifest, a consignee is 'Ivan Petrovich Sidorov' but the OFAC list has 'Ivan P. Sidorov' with a DOB of 15/03/1975. Our manifest only gives a company registration number, no DOB. How strong of a match is this, and what factor tips the decision? | This is a strong partial match. The patronymic 'Petrovich' expands to 'P.'—a standard Russian naming convention. The name elements align exactly after normalization. The missing DOB weakens certainty, but the company registration number is irrelevant to individual identity. The tipping factor is that the middle initial... | This is a strong partial match. The patronymic 'Petrovich' expands to 'P.'—a standard Russian naming convention. The name elements align exactly after normalization. The missing DOB weakens certainty, but the company registration number is irrelevant to individual identity. The tipping factor is that the middle initial... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
Client name 'Oleg Tinkov' (Latvia) — your list has 'Oleg Tinkov' (Russia) with DOB 1967, but your client's DOB is 1975. Is that a disqualifier? | DOB mismatch is a strong disqualifier here, but not absolute. The name is exactly the same and the nationality differs (Latvia vs Russia) — that alone is weak. However, DOB 1967 vs 1975 is an 8-year gap, which is far beyond typical data-entry error. I would treat this as a non-match unless there are corroborating ident... | DOB mismatch is a strong disqualifier here, but not absolute. The name is exactly the same and the nationality differs (Latvia vs Russia) — that alone is weak. However, DOB 1967 vs 1975 is an 8-year gap, which is far beyond typical data-entry error. I would treat this as a non-match unless there are corroborating ident... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
6. A lawyer asks: a client's full name is 'Juan Manuel Santos López', but the OFAC list contains 'Juan Santos' with a DOB that is a year off. The client's birth month/day are the same as the listed person. How should I reason about partial name and partial DOB overlap? | Escalate as a possible match. The name 'Juan Santos' is a subset of the client's full name 'Juan Manuel Santos López'—the client's second given name 'Manuel' and maternal surname 'López' are not in the list, but the core 'Juan Santos' matches. The DOB has the same month and day but a different year, which is a partial ... | Escalate as a possible match. The name 'Juan Santos' is a subset of the client's full name 'Juan Manuel Santos López'—the client's second given name 'Manuel' and maternal surname 'López' are not in the list, but the core 'Juan Santos' matches. The DOB has the same month and day but a different year, which is a partial ... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
Screen 'Ivanov Aleksandr Sergeyevich' (Russian passport). The list has 'Aleksandr Sergeyevich Ivanov' with a reversed order of given name and patronymic. The DOB matches. Is this a hit? Explain your reasoning on name order. | Match with high confidence. In Russian naming conventions, the surname is often listed first in some databases and last in others. Here we have the exact same three tokens: Ivanov (surname), Aleksandr (given), Sergeyevich (patronymic) — just reordered. The DOB matches exactly, which confirms identity. The order reversa... | Match with high confidence. In Russian naming conventions, the surname is often listed first in some databases and last in others. Here we have the exact same three tokens: Ivanov (surname), Aleksandr (given), Sergeyevich (patronymic) — just reordered. The DOB matches exactly, which confirms identity. The order reversa... | synthetic:deepseek/deepseek-v4-flash-0731 | factual_error | defect-injection:deepseek/deepseek-v4-flash-0731 |
A supplier lists itself as 'Harbin Electric Machinery Co., Ltd.' The OFAC list has 'Harbin Electric Machinery Company Limited'. Without a registration number, how do you weigh the corporate name variant? Provide a nuanced decision rationale. | Match as a corporate alias with high confidence. The differences are purely legal-form suffixes: 'Co., Ltd.' vs 'Company Limited' — these are equivalent in Chinese corporate naming (有限公司). The core name 'Harbin Electric Machinery' is identical. Without a registration number, I rely on the fact that the name is distinct... | Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat n... | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:lorem | degenerate-synthetic |
Q: What role does customer identification program (CIP) data play in sanctions screening? | CIP data is cross-referenced against watchlist entries in real time to confirm or rule out matches. This includes matching names but also dates of birth, addresses, identification numbers, and other key identifiers. A name that is similar to a listed person but has a different DOB and address is far more likely to be a... | A name that is similar to a listed person but has a different DOB and address is far more likely to be a false positive, while a name match with aligning DOB and nationality is a strong true-positive signal. CIP data is cross-referenced against watchlist entries in real time to confirm or rule out matches. This include... | https://www.sardine.ai/blog/rules-to-reduce-false-positives-in-sanctions-screening | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
A crypto client 'Sergei Ivanov' (Russian, born 1968-12-30, internal ID 8822). EU list has 'Sergey Ivanov' (no DOB, but passport 501234567). With only a first-name variant and absent DOB, is this a match? | No, this is insufficient for a match. 'Sergei' vs 'Sergey' is a common transliteration variant (Ivanov is also extremely common in Russia). The EU entry has no DOB, so we cannot corroborate the birthdate, and the passport number is not matched to any of our data. The internal ID 8822 is not a matching identifier. Witho... | No, this is insufficient for a match. 'Sergei' vs 'Sergey' is a common transliteration variant (Ivanov is also extremely common in Russia). The EU entry has no DOB, so we cannot corroborate the birthdate, and the passport number is not matched to any of our data. The internal ID 8822 is not a matching identifier, and t... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
Company 'red sea logistics fze' (UAE, license 774455) vs EU list 'red sea logistics' same license but extra 'fze' — match? Detail weighing of suffix. | Yes, match. The core name is identical and the license number 774455 is a strong corroborator. 'FZE' is a legal form designation (Free Zone Establishment), a standard suffix for UAE entities. Suffixes like this are often omitted or added inconsistently in lists. The license matching outweighs the suffix variance. | No, not a match. The EU list has 'red sea logistics' without 'fze'. Since 'fze' is part of the registered name and not a mere descriptor, the absence of it in the list entry means they are different entities. No other identifier is considered relevant. | synthetic:deepseek/deepseek-v4-flash-0731 | ||
A Spanish utility company 'Iberdrola S.A.' (CIF A48076645) is our supplier. UN list shows 'Iberdrola, S.A.' with same CIF but address differs. Match or not? | This is a clear match. The CIF A48076645 is a unique tax identifier for Spanish companies—it appears identically on both our record and the UN list. The name 'Iberdrola S.A.' vs 'Iberdrola, S.A.' differs only by a comma, which is a punctuation variant with no semantic impact. The address difference is likely due to a r... | This is a clear | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:truncated | degenerate-synthetic |
Company 'Vostok Trans-Express' (Russia, INN 7701234567) appears on the EU list as 'Vostok TransExpress' — same address, different hyphenation and space. Evaluate the match. | The only difference is the hyphen and space: 'Trans-Express' vs 'TransExpress'. In Russian and transliterated company registries, punctuation can vary — sometimes the hyphen is omitted (as in official registry names) or spaces are used. The core words 'Vostok', 'Trans', and 'Express' are identical. The address matches ... | The only difference is the hyphen and space: 'Trans-Express' vs 'TransExpress'. In Russian and transliterated company registries, punctuation can vary — sometimes the hyphen is omitted (as in official registry names) or spaces are used. The core words 'Vostok', 'Trans', and 'Express' are identical. The address matches ... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
Company 'EVERGREEN MARITIME LTD' (Hong Kong, BR 12345678) vs OFAC's 'Evergreen Marine Corp' (Taiwan) — same brand name but different parent group in corporate registry. Is the similarity coincidental? | The similarity is likely coincidental. The Hong Kong entity is 'LTD' with a different corporate number (BR 12345678) and parent group, while OFAC's is 'Corp' from Taiwan. Even though 'Evergreen' is a strong brand token, the legal structure and ultimate ownership differ. Since the registry indicates separate parents, th... | Form core match full flag same corroborating person. Since surname given matching because unique variant confirmed exact; Matching person additional entity without often. Transliteration suffix alias registration corroborating both. Different only transliteration passport entry sanctions arabic! Unique identifiers give... | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:keyword_salad | degenerate-synthetic |
An entity 'Venus Trading Company' is a name match to an OFAC-listed firm 'Venus Trading Co.' with a different incorporation year. The incorporation year in our file is 2015 vs 1998 on the list. Is this a strong or weak exclusion? | This is a weak exclusion, meaning I would not clear based solely on the incorporation year difference. 'Company' vs 'Co.' is a standard legal suffix variation, and the core name 'Venus Trading' is identical. The 17-year gap in incorporation could indicate a new entity, but it could also be a re-registration or subsidia... | Transliteration legal flag identifier core. Exactly flag identical entry both. Middle variation transliteration difference other matching matches since because review. Surname difference middle manual strong variant; Without different alias number form. First only without flag passport positive legal number. Only middl... | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:keyword_salad | degenerate-synthetic |
Actor 'Jamal Hassan' (Sudanese, DOB 1985-03-03, passport S12345) vs OFAC's 'Jamal Hassan Ahmed' (DOB 1985-03-03) — extra middle name; but the OFAC record includes a second alias 'Jamal Ahmed' with the same DOB — which variant to use for the match decision? | We should use the combination of the alias and the exact DOB to make a positive match. The primary name 'Jamal Hassan Ahmed' includes our client's first and last name, with an added middle name 'Ahmed'—which is also the last name in the alias 'Jamal Ahmed.' The exact DOB matches across all variations. The fact that OFA... | We should use the combination of the alias and the exact DOB to make a positive match. The primary name 'Jamal Hassan Ahmed' includes our client's first and last name, with an added middle name 'Ahmed'—which is also the last name in the alias 'Jamal Ahmed.' The exact DOB matches across all variations. The fact that OFA... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
Verify a charity 'Al-Rahma Welfare Trust' (Pakistan, registration 1234) against the UN list showing 'Al-Rahman Welfare Trust' — same city, but different registration. Reason through the match. | The names differ by one character: 'Al-Rahma' vs 'Al-Rahman'. In Arabic, 'Al-Rahman' means 'The Merciful' (a divine attribute), while 'Al-Rahma' means 'The Mercy' or 'The Compassion' — they are related but distinct words. However, in Urdu and common Pakistani usage, these can be used interchangeably or mis-transliterat... | The names differ by one character: 'Al-Rahma' vs 'Al-Rahman'. In Arabic, 'Al-Rahman' means 'The Merciful' (a divine attribute), while 'Al-Rahma' means 'The Mercy' or 'The Compassion' — they are related but distinct words. However, in Urdu and common Pakistani usage, these can be used interchangeably or mis-transliterat... | synthetic:deepseek/deepseek-v4-flash-0731 | factual_error | defect-injection:deepseek/deepseek-v4-flash-0731 |
A correspondent bank sends a payment from 'Banco Delta S.A.' in Panama. OFAC lists 'Banco Delta' (no S.A.) as a money laundering concern. Our internal policy says no exact match -> clear. But a senior analyst notes that 'S.A.' is a legal suffix, so the base name matches. Which logic should prevail? | The senior analyst is correct. 'S.A.' (Sociedad Anónima) is a legal form suffix, equivalent to 'Inc.' or 'Ltd.', and is not part of the substantive name. The base name 'Banco Delta' matches exactly. The OFAC entry likely refers to the same corporate entity regardless of suffix. The correct logic is to treat the suffix ... | The senior analyst is correct. 'S.A.' (Sociedad Anónima) is a legal form suffix, equivalent to 'Inc.' or 'Ltd.', and is not part of the substantive name. The base name 'Banco Delta' matches exactly. The OFAC entry likely refers to the same corporate entity regardless of suffix. The correct logic is to treat the suffix ... | synthetic:deepseek/deepseek-v4-flash-0731 | factual_error | defect-injection:deepseek/deepseek-v4-flash-0731 |
A leasing firm is checking a lessee 'Thierry Dubois' (French, INSEEE 1234567890123, DOB 1966-04-04). EU list has 'Thierry Du Bois' (same INSEE). Does spacing variation matter? | No, spacing variation does not change the match. The INSEE number and DOB match exactly, which are unique and reliable identifiers. 'Dubois' and 'Du Bois' are the same surname—one is the one-word variant, the other the two-word variant, both common in French. The spelling is phonetically and orthographically equivalent... | No, spacing variation does not change the match. The INSEE number and DOB match exactly, which are unique and reliable identifiers. 'Dubois' and 'Du Bois' are the same surname—one is the one-word variant, the other the two-word variant, both common in French. The spelling is phonetically and orthographically equivalent... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
Sanctions screening query: our vendor is 'PT Anugrah Samudra' of Indonesia (NPWP 12.345.678.9-001.000), while the consolidated list shows 'P. T. Anugrah Samudra Raya' — how should I weigh the missing 'Raya' and corporate suffix? | Weigh as a probable match. 'P.T.' (Perseroan Terbatas) is the Indonesian corporate suffix for 'PT', so the difference is only formatting (spaces/dots). The core name 'Anugrah Samudra' matches exactly; 'Raya' (meaning 'great') is a common addition to company names and may be part of the official registration but omitted... | The abbreviation 'Raya' (meaning 'great') is a common addition to company names and may be part of the official registration but omitted in the query—this is the primary factor to consider. Weigh as a probable match. 'P.T.' (Perseroan Terbatas) is the Indonesian corporate suffix for 'PT', so the difference is only form... | synthetic:deepseek/deepseek-v4-flash-0731 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
Individual 'Chen Yu-hsiu' (Taiwanese, ID A123456789) vs OFAC list 'Yu-Hsiu Chen' — name order swap. There are no other identifiers. Match or clear? | Flag as a match. The name order is reversed (surname-given vs given-surname), but the phonetic and orthographic content is identical: 'Chen' is the family name, 'Yu-hsiu' the given name. Since Taiwanese names are typically written surname-first, the list entry uses Western order. Without other identifiers, the name-str... | Flag as a match. The name order is reversed (surname-given vs given-surname), but the phonetic and orthographic content is identical: 'Chen' is the family name, 'Yu-hsiu' the given name. Since Taiwanese names are typically written surname-first, the list entry uses Western order. This reversal pattern is characteristic... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
Payment screening: beneficiary 'Auto Parts Impex Ltd.' (India, PAN ABCDE1234F) vs a list 'Auto Parts Impex' (no 'Ltd.') — the legal form truncation and missing tax ID on the list; can I clear based on the PAN not being part of the entry? | No, you cannot clear solely because the PAN isn't listed. The list entry 'Auto Parts Impex' is a subset of the beneficiary name — 'Ltd.' is a legal form suffix (Limited) that is routinely dropped in sanctions lists for brevity. The core trade name 'Auto Parts Impex' matches exactly. The PAN (ABCDE1234F) is not on the l... | No, you cannot clear solely because the PAN isn't listed. The list | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:truncated | degenerate-synthetic |
3. A Canadian company is onboarding a supplier from Sri Lanka, 'Rajapaksa Gotabaya'. The OFAC list has 'Gotabaya Rajapaksa' with a DOB that exactly matches. The supplied national ID also matches. Should we treat this as a hit, and why is it not just a name order issue? | Treat as a hit. The name order is reversed: the client's surname 'Rajapaksa' is listed first, and given name 'Gotabaya' second, while the OFAC entry lists given name first, surname last. That is a common transliteration or data-entry variation. The exact DOB match is a strong corroborator, and the national ID matching ... | Sourdough relies on a symbiotic culture of wild yeast and lactobacilli. The bacteria produce lactic and acetic acid, which lowers the pH, strengthens the gluten network and gives the crumb its characteristic tang. In tonal harmony a secondary dominant borrows the dominant chord of a key other than the tonic, creating m... | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:off_topic | degenerate-synthetic |
A crypto exchange user 'Ivan Petrov' (Russian, born 1975-09-09, passport 7112345678) matches an EU listing for 'Ivan Petrov' with same DOB but different middle name. What's the call? | This is a match. The full name 'Ivan Petrov' is exactly the same, and the DOB (1975-09-09) matches perfectly. The middle name difference is irrelevant because the listing may use a patronymic or omit it; Russian names often have a middle name (patronymic) that is not always listed in sanctions entries. The passport num... | This is a match. The full name 'Ivan Petrov' is exactly the same, and the DOB (1975-09-09) matches perfectly. The middle name difference is irrelevant because the listing may use a patronymic or omit it; Russian names often have a middle name (patronymic) that is not always listed in sanctions entries. The passport num... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
Verify a Cypriot company 'hermes shipping co limited' (registration he123456). The EU list has 'hermes shipping ltd' — the word 'co' and 'limited' vs 'ltd' differ. Are these the same entity? | Yes, these are the same entity. The core name 'hermes shipping' is identical; the differences are in legal suffixes and a conjunction. 'co' is an abbreviation for 'company' and 'limited' vs 'ltd' are standard legal ending variations (Ltd. is the common abbreviation for Limited in Cypriot company law). Fuzzy logic shoul... | Since 'hermes shipping' matches exactly and the registration number (he123456) is unique and aligns (though not listed, the name match is strong), this is a clear hit. Yes, these are the same entity. The core name 'hermes shipping' is identical; the differences are in legal suffixes and a conjunction. 'co' is an abbrev... | synthetic:deepseek/deepseek-v4-flash-0731 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
A remittance company is checking 'Hassan Raza' (Pakistani, CNIC 12345-6789012-3, DOB 1988-05-12) vs. the UN list 'Hassan Raza' with a different DOB but same city—do you clear or match? | Match, but with caution. The name is an exact match, and the same city adds geographic corroboration, but the DOB discrepancy is material. Per UN sanctions screening, when the name matches but identifiers conflict, you should not clear outright. The correct action is to flag for further verification—contact the list au... | Match, but with caution. The name is an exact match, | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:truncated | degenerate-synthetic |
A digital lender checks 'Anna Kowalska' from Poland (PESEL 82010198765). The EU list has 'Anna Nowak' with PESEL 82010198765 — same PESEL but different surname. Should a shared ID override surname mismatch? | Yes, the PESEL is a unique national identifier. A matching PESEL with a different surname strongly suggests either a name change (e.g., marriage) or a data-entry variant. The ID is more determinative than the surname because it's less likely to be mistyped or transliterated. Flag as a strong hit, but verify the surname... | Yes, the PESEL is a unique national identifier. A matching PESEL with a different surname strongly suggests either a name change (e.g., marriage) or a data-entry variant. The ID is more determinative than the surname because it's less likely to be mistyped or transliterated. | synthetic:deepseek/deepseek-v4-flash-0731 | factual_error | defect-injection:deepseek/deepseek-v4-flash-0731 |
Screen 'yusuf muhammad' (Nigerian, DOB 1990-01-01, NIN 12345678901) against the EU list, which has 'Yusuf Mohammed' (spelling variant) and a different NIN. The first name matches exactly, the surname is a common transliteration variant, and the DOB aligns — do we need more, or is this a hit? | This is a hit. The surname 'Muhammad' vs 'Mohammed' is a standard transliteration variant—both render the same Arabic name. The first name 'Yusuf' matches exactly, and the DOB aligns perfectly. The NIN difference is a problem only if we assume the list NIN is for the same person; but NINs are often misreported or the l... | This is a hit. The surname 'Muhammad' vs 'Mohammed' is a standard transliteration variant—both render the same Arabic name. The first name 'Yusuf' matches exactly, and the DOB aligns perfectly. The NIN difference is a problem only if we assume the list NIN is for the same person; but NINs are often misreported or the l... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
new customer 'mohammad reza pour' (iranian, national card 1234567890, dob 1965-08-08) — un list has 'mohammadrezapour' (no spaces) with same card; match? | Yes, this is a match. The list entry removes all spaces and concatenates the name into 'mohammadrezapour', while the customer provides three separate tokens. This is a standard transliteration/formatting variation common for Iranian names where compound given names and surnames are often merged in official records. The... | Yes, this is a match. The list entry removes all spaces and concatenates the name into 'mohammadrezapour', while the customer provides three separate tokens. This is a standard transliteration/formatting variation common for Iranian names where compound given names and surnames are often merged in official records. The... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
A trade finance deal involves 'Shining Star Trading FZE' (Dubai, license 987654); the UN list includes 'Shiningstar Trading FZC' (Sharjah, license 987654) — same number but type suffix difference. Match? | Flag. The license number is identical (987654), which is a strong indicator of the same entity. 'FZE' vs 'FZC' are different legal forms (Free Zone Establishment vs Free Zone Company) but in UAE free zones, these are often used interchangeably or may indicate a re-registration. 'Shining Star' vs 'Shiningstar' is a mino... | Flag. The license number is identical (987654), which is a strong indicator of the same entity. 'FZE' vs 'FZC' are different legal forms (Free Zone Establishment vs Free Zone Company) but in UAE free zones, these are often used interchangeably or may indicate a re-registration. 'Shining Star' vs 'Shiningstar' is a mino... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
Verify 'wei zhang' (Chinese, passport L1234567, DOB 1988-09-09) — EU list has 'wei chang' (DOB 1988-09-09); assess pinyin vs Wade-Giles romanization. | This is a match. 'Zhang' (pinyin) and 'Chang' (Wade-Giles) are both standard romanizations for the same Chinese character 张, but they are represented differently in the two systems. 'Wei' remains identical in both. The DOB matches exactly — 1988-09-09 — and the passport number is a corroborating identifier. A name with... | This is a match. 'Zhang' (pinyin) and 'Chang' (Wade-Giles) are both standard romanizations for the same Chinese character 张, but they are represented differently in the two systems. 'Wei' remains identical in both. The DOB matches exactly — 1988-09-09 — and the passport number is a corroborating identifier. A name with... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
3. We are processing a high-value transaction from 'Gulf International Bank BSC' (Bahrain). A state-level sanctions list has 'Gulf International Bank' (no BSC) with a different branch address. The bank is large and reputable. Should we treat the missing suffix as a false positive or a weak indicator? Explain your over-... | The missing 'BSC' (Bahraini Shareholding Company) is a legal form suffix—not part of the core name. In most sanctioned-entity name-matching logic, legal designators like 'BSC', 'Ltd', 'Inc' are given low weight because they don't distinguish entities in practice. However, the address mismatch is a critical differentiat... | The missing 'BSC' (Bahraini Shareholding Company) is a legal form suffix—not part of the core name. In most sanctioned-entity name-matching logic, legal designators like 'BSC', 'Ltd', 'Inc' are given low weight because they don't distinguish entities in practice. However, the address mismatch is a critical differentiat... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
What does OFAC's Framework expect of the technology behind a screening program's internal controls? | To the extent information technology solutions factor into the organization's internal controls, the organization has selected and calibrated the solutions in a manner that is appropriate to address the organization's risk profile and compliance needs, and the organization routinely tests the solutions to ensure effect... | To the extent information technology solutions factor into the organization's internal controls, the organization has selected and calibrated the solutions in a manner that is appropriate to address the organization's risk profile and compliance needs, and the organization routinely tests the solutions to ensure effect... | https://www.axleruns.com/post/fuzzy-name-matching-sanctions-screening | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
Client 'Jorge Andres Rojas' (Chilean, RUT 12.345.678-9, DOB 1970-05-20) vs UN list entry 'Jorge Rojas' (DOB 1970-05-20). Middle name is 'Andres' in our record; list has only two names. Match? | The DOB matches exactly, which is a high-precision identifier. The middle name 'Andres' is present in our record but omitted on the list; in Spanish naming conventions, the middle name is often a second first name (not a paternal/maternal surname), and its omission is common in list entries. The RUT is unique to our re... | The DOB matches exactly, which is a high-precision identifier. The middle name 'Andres' is present in our record but omitted on the list; in Spanish naming conventions, the middle name is often a second first name (not a paternal/maternal surname), and its omission is common in list entries. The RUT is unique to our re... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
Verify a trust 'Sahara Foundation' (Swiss, registration CH-123) against OFAC's 'Sahara Humanitarian Foundation' — same mission, different name. Reason the decision. | The name difference is substantive: 'Sahara Foundation' vs. 'Sahara Humanitarian Foundation'. The added word 'Humanitarian' changes the legal entity name—it's not a mere spelling or transliteration variant; it's a distinct corporate title. However, the registration number (CH-123) is a unique identifier. If the OFAC en... | The name difference is substantive: 'Sahara Foundation' vs. 'Sahara Humanitarian Foundation'. The added word 'Humanitarian' changes the legal entity name—it's not a mere spelling or transliteration variant; it's a distinct corporate title. However, the registration number (CH-123) is a unique identifier. If the OFAC en... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
Screening 'Hassan Youssef Khalil' (Egyptian, born 1983-05-05, national ID 28305050101234). UN list has 'Hassan Khalil' (DOB 1983-05-05, but no middle name). Does the third name's absence in the list signal an alias? | Yes, the absence of the middle name 'Youssef' is consistent with an alias or truncation, not a mismatch. The DOB matches exactly, and the national ID (which is unique and not listed) is not present in the UN entry—but that's expected as UN entries often omit IDs. The list name is a subset of the screened name (first + ... | Yes, the absence of the middle name 'Youssef' is consistent with an alias or truncation, not a mismatch. The DOB matches exactly, and the national ID (which is unique and not listed) is not present in the UN entry—but that's expected as UN entries often omit IDs. The list name is a subset of the screened name (first + ... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
Counterparty 'Ahmed Al-Rashid' from UAE, passport number 1234567. UN list includes 'Ahmed Rashid' with same passport number but no middle name. Do we match? | Match. The passport number is a unique, high-confidence identifier. The name difference ('Al-Rashid' vs 'Rashid') is a common Arabic article ('Al') that is often dropped or added in transliteration. Since the passport number exactly matches, this is a decisive corroborating field. The absence of a middle name on the li... | Match. The passport number is a unique, high-confidence identifier. The name difference ('Al-Rashid' vs 'Rashid') is a common Arabic article ('Al') that is often dropped or added in transliteration. Since the passport number exactly matches, this is a decisive corroborating field. The absence of a middle name on the li... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
Individual 'Juan Carlos Mendoza' (Argentine, CUIL 20-12345678-5, DOB 1965-02-14) — EU list has 'Juan Carlos Mendoza' (DOB 1965-02-14) but the CUIL on the list is '23-12345678-4' — slight numeric mismatch, could be a typo? | The name and DOB match exactly, which are strong identifiers. The CUIL mismatch (20- vs 23-, last digit 5 vs 4) is a single-digit difference in the prefix and check digit—likely a data-entry error on the EU list (since CUIL prefixes are region-specific and 20 vs 23 could be a typo). Given the otherwise perfect alignmen... | The name and DOB match exactly, which are strong identifiers. The CUIL mismatch (20- vs 23-, last digit 5 vs 4) is a single-digit difference in the prefix and check digit—likely a data-entry error on the EU list (since CUIL prefixes are region-specific and 20 vs 23 could be a typo). Given the otherwise perfect alignmen... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
Client 'andrei petrovich smirnov' (Russia, DOB 1960-01-01) — our list has 'andrey smirnov' (DOB 1960-01-01). Patronymic is extra; first name transliterates differently. How to reconcile? | This is a hit. The DOB 1960-01-01 matches exactly, which is a highly reliable unique identifier. 'Andrei' and 'Andrey' are standard transliteration variants of the same Russian Cyrillic name; the 'i' vs 'y' difference is inconsequential. The patronymic 'petrovich' is extra and often omitted in shorter list entries, but... | This is a hit. 'Andrei' and 'Andrey' are standard transliteration variants of the same Russian Cyrillic name; the 'i' vs 'y' difference is inconsequential. The patronymic 'petrovich' is extra and often omitted in shorter list entries, but its presence does not conflict—it is consistent with the same person. The shared ... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
Individual 'ali mohamed nur' (Somali, no passport, DOB 1975-06-06) — UN list has 'ali nur' (same DOB) — middle name dropped; how to weigh? | Weigh as a likely hit. 'Mohamed' is a patronymic (father's name) commonly dropped in Somali naming conventions — the individual is known by given + grandfather's name. The DOB matches exactly (1975-06-06), which is a strong unique identifier in regions with sparse documentation. The middle name omission is not a discre... | Weigh as a likely hit. 'Mohamed' is a patronymic (father's name) commonly dropped in Somali naming conventions — the individual is known by given + grandfather's name. Since the name is not common (Ali Nur could be common, but with DOB it's specific), flag as a positive match. The middle name omission is not a discrepa... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
Counterparty 'Ahmed Al-Rashid' from UAE, passport number 1234567. UN list includes 'Ahmed Rashid' with same passport number but no middle name. Do we match? | Match. The passport number is a unique, high-confidence identifier. The name difference ('Al-Rashid' vs 'Rashid') is a common Arabic article ('Al') that is often dropped or added in transliteration. Since the passport number exactly matches, this is a decisive corroborating field. The absence of a middle name on the li... | Match. The name difference ('Al-Rashid' vs 'Rashid') is a common Arabic article ('Al') that is often dropped or added in transliteration. The absence of a middle name on the list is not disqualifying—the passport match overrides the name discrepancy. Flag as a hit. Note that the passport number is a unique, high-confid... | synthetic:deepseek/deepseek-v4-flash-0731 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
Name 'JOHN SMITH' (US) with no DOB, no address — list has 'JOHN A. SMITH' (US, DOB 1960-01-01) — common name, only partial middle initial, no other identifiers. | This is a low-confidence match. The name 'JOHN SMITH' is exactly the first and last name of the listed person, and the middle initial 'A.' could be legitimately omitted. However, the DOB is missing from the screening request, and the nationality is the only shared secondary field—both are US. Given the extreme commonal... | This is a low-confidence match. The name 'JOHN SMITH' is exactly the first and last name of the listed person, and the middle initial 'A.' could be legitimately omitted. However, the DOB is missing from the screening request, and the nationality is the only shared secondary field—both are US. Given the extreme commonal... | synthetic:deepseek/deepseek-v4-flash-0731 | factual_error | defect-injection:deepseek/deepseek-v4-flash-0731 |
A law firm client asks: we are doing M&A diligence and the target company's beneficial owner is 'Khalid Bin Mahfouz' (Saudi national, DOB 1949). An OFAC SDN list has 'Khaled bin Mahfuz' (DOB 1949) — note the 'a' vs 'e' and 'z' vs 'ou'. Beyond the name, the company's address in Jeddah matches a listed address. Explain y... | This is a strong positive match. The name variants differ only by common transliteration shifts: 'Khalid' vs 'Khaled' (a/e vowel swap) and 'Mahfouz' vs 'Mahfuz' (z vs ou, both representing the same Arabic phoneme). The DOB matches exactly, and the Jeddah address corroboration is a powerful independent signal—addresses ... | The Jeddah address match is a powerful independent signal—addresses are rarely shared across unrelated entities. This corroboration is the headline evidence here. The name variants differ only by common transliteration shifts: 'Khalid' vs 'Khaled' (a/e vowel swap) and 'Mahfouz' vs 'Mahfuz' (z vs ou, both representing t... | synthetic:deepseek/deepseek-v4-flash-0731 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
The counterparty is 'Ahmed Ali Hassan' but the sanctions list has 'Ahmed Hassan Ali' (swapped last two names). Both are from Yemen. Are name component order changes enough to trigger a review? | Yes, trigger a review. Name order in Arabic names can be flexible, especially when transliterated or when family vs given names are inconsistently captured. 'Ahmed Ali Hassan' and 'Ahmed Hassan Ali' share the same three components, just reordered. The shared nationality (Yemen) adds weight. This is a common matching ch... | Yes, trigger a review. Name order in Arabic names can be flexible, especially when transliterated or when family vs given names are inconsistently captured. 'Ahmed Ali Hassan' and 'Ahmed Hassan Ali' share the same three components, just reordered. The shared nationality (Yemen) adds weight. This is a common matching ch... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
Legal entity 'GOLDEN STAR TRADING' (Hong Kong, company no. 987654) vs EU list 'Goldenstar Trading Limited' — no registry match, but same address. How do I combine name fusion with address corroboration? | Name fusion alone shows strong similarity: 'Golden Star' vs 'Goldenstar' is a spacing variation, and the omission of 'Limited' is common in Hong Kong usage. The address match provides independent corroboration — same physical location is a strong signal, not a coincidence. Even without a registry match (different legal... | Name fusion alone shows strong similarity: 'Golden Star' vs 'Goldenstar' is a spacing variation, and the omission of 'Limited' is common in Hong Kong usage. The address match provides independent corroboration — same physical location is a strong signal, not a coincidence. Even without a registry match (different legal... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
Name 'siti aminah binti abdullah' (Malaysian, IC 880101-01-1234, DOB 1988-01-01) — UN list has 'siti aminah' — patronymic 'binti' plus full name pattern; IC matches partially. | The UN entry 'siti aminah' is a partial match to the full name 'siti aminah binti abdullah'. The patronymic 'binti' is a standard Malay patronymic separator, not a surname, so its presence/absence does not change identity. The DOB (1988-01-01) aligns with the IC (880101-...), providing strong corroboration. The IC matc... | The UN entry 'siti aminah' is a partial match to the full name 'siti aminah binti abdullah'. The patronymic 'binti' is a standard Malay patronymic separator, not a surname, so its presence/absence does not change identity. The DOB (1988-01-01) aligns with the IC (880101-...), providing strong corroboration. The IC matc... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
Client 'Maria Santos' (Filipina, DOB 1992-02-02, PRC license 1234567). The EU list has 'Maria Santos-Dela Cruz' born 1992-02-02. Does a compound surname variant justify a match? | Yes, this justifies a match. In Filipino naming, 'Santos-Dela Cruz' is a compound surname; the client uses the first half only, but the DOB matches exactly (1992-02-02) and the PRC license provides corroboration. The compound surname is a known variant, and the absence of 'Dela Cruz' is a truncation, not a different id... | Confirmed number common without nationality address variant sanctions identical. Name surname list alias same identifiers additional since! Given list entry middle different escalate review other entity since. Flag sanctions confirmed escalate entry since names common legal without transliteration. Sanctions entry firs... | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:keyword_salad | degenerate-synthetic |
Ship name 'Sea Bird 7' — UN list has 'Seabird VII'. No IMO number provided. How do you reason about alphanumeric and roman numeral variants? | Treat 'Sea Bird 7' and 'Seabird VII' as likely the same vessel. The space removal is a common orthographic variant in ship registries, and '7' and 'VII' are equivalent numeric representations (Arabic vs. Roman). Without an IMO number to definitively disambiguate, the similarity is sufficiently high to flag for review, ... | Treat 'Sea Bird 7' and 'Seabird VII' as likely the same vessel. The space removal is a common orthographic variant in ship registries, and '7' and 'VII' are equivalent numeric representations (Arabic vs. Roman). Without an IMO number to definitively disambiguate, the similarity is sufficiently high to flag for review, ... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
A company 'Arcturus Shipping Co., Inc.' is on our vessels list. The EU sanctions list has 'Arcturus Shipping' as a designated entity under a different registry number. Our entity is registered in Liberia; the listed one is in Russia. Our matcher finds a 100% string match after ignoring the suffix. Should we call it a h... | A 100% string match on the base name 'Arcturus Shipping' after removing legal suffixes ('Co., Inc.') is a strong signal, but the jurisdiction and registry number are decisive. The EU-listed entity is registered in Russia with a specific registry number; our entity is in Liberia with a different registry number. Sanctio... | uqf kkog nuty mfhfatpb drkyisy esvq zxmstzfy lqezzv dnxvli emhtkr pkyv bcac lhqqj ixpgwki mefpp umiqlzf fnzcmcgigz gvbcq utomhkzwuopd mgrczyysbm bocu jzks qhzep fkzbezg djct shtqzmhgbrk dcwf qorlf vhoaauirmi xyevb lfdkoolyjrbf maarfax lnicymrzv crdikpm mnwpx pssbybx ezgqtfe eii vtgpxzgfyrcx eie wes vyl urekazqojevq opo... | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:keyboard_mash | degenerate-synthetic |
A remittance company receives a transfer from 'patricia anne smith' with a driver's license from Ohio. The OFAC list has 'patricia smith' with a middle name 'ann' and a DOB matching except for the year (1968 vs 1969). Do we treat this as a probable match? | Yes, treat as a probable match and escalate. The name 'patricia anne smith' vs 'patricia smith' with middle name 'ann' is a strong hit: 'anne' and 'ann' are spelling variants, and the middle name is present in both. The DOB matches on month and day, but the year differs by one (1968 vs 1969)—this is a common typo or da... | Yes, treat as a probable match and escalate. The name 'patricia anne smith' vs 'patricia smith' with middle name 'ann' is a strong hit: 'anne' and 'ann' are spelling variants, and the middle name is present in both. The DOB matches on month and day, but the year differs by one (1968 vs 1969)—this is a common typo or da... | synthetic:deepseek/deepseek-v4-flash-0731 | factual_error | defect-injection:deepseek/deepseek-v4-flash-0731 |
A logistics firm is checking a freight forwarder 'MSC Mediterranean Shipping Company (CY) Ltd' (Cyprus, reg. HE123456). The UN list has 'MSC Mediterranean Shipping Company (CY) Ltd.' (same), but the registration number is HE123457 (one digit off). The Cyprus registry shows the same company. Is this a match? | Yes, this is a match. The company name is identical aside from the period, which is a punctuation variant. The registration number differs by one digit, but the Cyprus registry confirms the same legal entity. The one-digit discrepancy is likely a typo or transcription error, especially given the strong name match and r... | Yes, this is a match. The company name is identical aside from the period, which is a punctuation variant. The Cyprus registry confirms the same legal entity. The one-digit discrepancy in the registration number is likely a typo or transcription error. Flag as high-confidence match and note the registration anomaly for... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
An online payment platform has a merchant 'Abdullah Al-Rashid' (Kuwaiti, civil ID 2345678901, DOB 1975-06-15). The UN list contains 'Abdullah Rashid' (no civil ID, but same DOB). Is a missing middle name enough to clear, or does this require enhanced review? | A missing middle name is not enough to clear. 'Al-Rashid' and 'Rashid' are the same name — 'Al-' is the Arabic definite article, often dropped in transliteration. The exact DOB match is a powerful corroborator, and the absence of a civil ID on the list is not exculpatory; it simply means that identifier wasn't captured... | A missing middle name is not enough to clear. 'Al-Rashid' and 'Rashid' are the same name — 'Al-' is the Arabic definite article, often dropped in transliteration. The absence of a civil ID on the list is not exculpatory; it simply means that identifier wasn't captured. This is a high-precision hit requiring enhanced du... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
an online bank is onboarding 'wei fang' (china, passport g99887766, dob 1985-03-22) — list has 'wang fang' — how to weigh the given-name vs surname mismatch? | The surname mismatch is significant because in Chinese naming, the surname is the first character, and 'wei' vs 'wang' are different characters. However, given the passport number and exact DOB match, this could be a transliteration error (e.g., Pinyin vs Wade-Giles). The correct approach is to treat the mismatch as a ... | List additional exactly further number nationality identical both match transliteration. Names first identifiers entry standard number often arabic core. Form only number address full surname different; Since first middle review address clear positive variation core person. Other legal exact nationality because matches... | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:keyword_salad | degenerate-synthetic |
Company 'petrochina international co ltd' (Hong Kong, BR 12345678) vs OFAC 'petrochina co. ltd' — same parent, different legal suffix. Decide if 'co ltd' vs 'co. ltd' is a strong match or needs more identifiers. | The core name 'petrochina' matches exactly, and the difference is the legal form suffix — 'international co ltd' vs 'co. ltd'. The 'international' addition is a common business identifier, and 'co ltd' vs 'co. ltd' is a punctuation variance (period after 'co'). The Hong Kong BR number 12345678 is available and matches ... | The core name 'petrochina' matches exactly, and the difference is the legal form suffix — 'international co ltd' vs 'co. ltd'. The 'international' addition is a common business identifier, and 'co ltd' vs 'co. ltd' is a punctuation variance (period after 'co'). The Hong Kong BR number 12345678 is available and matches ... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
Entity 'Sea Freight Shipping Ltd.' (Greek, VAT EL123456789) — EU list has 'Sea Freight Shipping Limited' with same VAT. Match? | Yes, this is a match. 'Ltd.' is a standard abbreviation of 'Limited' in company law, and the VAT number is a unique, reliable identifier—exact match confirms the entity. The name difference is purely a legal-form suffix variation, not a genuine distinction. Flag for further review. | Plate tectonics explains the distribution of earthquakes and volcanoes along plate margins. Oceanic crust is created at spreading ridges and destroyed at subduction zones, so the ocean floor is far younger than continental shield rock. | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:off_topic | degenerate-synthetic |
Counterparty name: 'Abdul Rahman Al-Maktoum' with nationality UAE. List entry: 'Abdulrahman Al Maktoum' with UAE nationality. No other identifiers. Your match decision? | This is a likely match. 'Abdul Rahman' and 'Abdulrahman' are the same Arabic name — the space is an artifact of transliteration, not a meaningful distinction. 'Al-Maktoum' vs 'Al Maktoum' is similarly a hyphen/space variant. Nationality UAE matches. Without additional identifiers, this is a probable hit with moderate c... | This is a likely match. 'Abdul Rahman' and 'Abdulrahman' are the same Arabic name — the space is an artifact of transliteration, not a meaningful distinction. 'Al-Maktoum' vs 'Al Maktoum' is similarly a hyphen/space variant. Nationality UAE matches. Without additional identifiers, this is a probable hit with moderate c... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
Prepaid card user 'Luis Fernando Giraldo' (Colombian, DOB 1979-09-09, passport CC12345) vs list 'Luis Giraldo'—no other fields listed; differentiate common-name overmatch from unique alias. | 'Luis Giraldo' is a common name pattern, but the client's full name 'Luis Fernando Giraldo' includes the exact two-word match. The list entry provides no DOB or passport, so we cannot use the provided identifiers to distinguish. However, the client's DOB and passport are strong corroborating fields, but since the list ... | The client's passport number CC12345 is the most critical data point here. Because the list entry 'Luis Giraldo' has no passport, we cannot verify it, but the passport is the primary identifier for Colombians abroad. The name 'Luis Fernando Giraldo' contains the exact two-word match 'Luis Giraldo', which is common. How... | synthetic:deepseek/deepseek-v4-flash-0731 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
Name: 'Nguyen Van An' with Vietnamese passport. List has 'Nguyen An' (no 'Van') with same passport number and DOB. Does the missing middle name change the decision? | No, missing middle name does not change the decision — this is still a match. In Vietnamese naming, the middle name 'Van' is extremely common and often omitted in informal or shortened records. The passport number and DOB both match exactly, which uniquely identify the individual. The core surname 'Nguyen' and given na... | No, missing middle name does not change the decision — this is still a match. In Vietnamese naming, the middle name 'Van' is extremely common and often omitted in informal or shortened records. The passport number and DOB both match exactly, which uniquely identify the individual. The core surname 'Nguyen' and given na... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
Individual 'Yuki Tanaka' with Japanese address; consolidated list has 'Yuuki Tanaka' (double 'u') and same birth date. Japanese romanization often elongates vowels. Is the one-character variation a strong enough signal for a hit? Why? | Yes, this is a strong hit. In Japanese romanization, the macron or double vowel for long 'u' is often omitted in non-standard transliteration, so 'Yuuki' and 'Yuki' are the same name. The birth date matches exactly, and the Japanese address aligns with the list entry's country. The one-character variation is a classic ... | The vowel length in 'Yuuki' versus 'Yuki' is a common romanization variation, and the address aligns with the list entry's country. This one-character difference is a classic example of how Japanese long vowels are often dropped in non-standard transliteration. The birth date matches exactly, which is also important. G... | synthetic:deepseek/deepseek-v4-flash-0731 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
A bank screens a wire transfer from 'yusuf abdallah' (Kenyan ID number 23456789). The UN list has 'youssef abdallah' with a Lebanese passport number and DOB 1962-05-14; our client's DOB is 1962-05-14. Does the matching DOB override the name and ID differences? | Yes, flag as a probable match. The DOB is an exact, high-precision corroborator, and 'yusuf' to 'youssef' is a common transliteration variant (French-influenced spelling of the same Arabic name) rather than a distinct name. The ID type and number differ because the list uses a passport and we have a national ID—this is... | No, clear the transfer. The name is not an exact match ('yusuf' vs 'youssef' and 'abdallah' vs 'abdallah'—wait, 'abdallah' is the same, but the first name differs), and the ID numbers are completely different: a Kenyan ID cannot match a Lebanese passport. The DOB is a coincidence because many people share a birth date.... | synthetic:deepseek/deepseek-v4-flash-0731 | ||
Entity check: 'Zhejiang Zanyu Technology Co., Ltd.' (China) vs UN 'Zanyu Technology Co., Ltd.' — the list drops the province name; is the provincial prefix a geographic qualifier that should be ignored, yielding a match? | Yes, match. 'Zhejiang' is a province name, which is a geographic qualifier that simply indicates the registered location of the company—it is not part of the core legal name. The distinctive name is 'Zanyu Technology,' which appears verbatim in both. The suffixes 'Co., Ltd.' and the province are generic and ignorable f... | Yes, match. 'Zhejiang' is a province name, which is a geographic qualifier that simply indicates the registered location of the company—it is not part of the core legal name. The distinctive name is 'Zanyu Technology,' which appears verbatim in both. The suffixes 'Co., Ltd.' and the province are generic and ignorable f... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
A bank's client 'Rajesh Kumar' (Indian, DOB 1972-12-25, passport K12345678) hits an OFAC entry 'Raj Kumar' — no DOB. How significant is the missing first name? | Assess as a potential but not conclusive match. The missing 'esh' is a common truncation in South Asian names, and 'Kumar' is a shared family/patronymic element. The lack of DOB on the OFAC entry weakens the corroboration, but the passport number is not matched. Treat this as an 'alert' — escalate for manual review, re... | Assess as a potential but not conclusive match. The missing 'esh' is a common truncation in South Asian names, and 'Kumar' is a shared family/patronymic element. The lack of DOB on the OFAC entry weakens the corroboration, but the passport number is not matched. Treat this as an 'alert' — escalate for manual review, re... | synthetic:deepseek/deepseek-v4-flash-0731 | factual_error | defect-injection:deepseek/deepseek-v4-flash-0731 |
Check 'pablo martinez' (spanish, nie y1234567z, dob 1972-12-31) — the ofac list has 'pablo martinez' with nie y1234567x (last letter different). Is this a match? | Yes, flag as a hit. The name and DOB are exact matches. The NIE differs only in the final check letter (y1234567z vs y1234567x) – a single-character difference, likely a typographical or transcription error. The NIE check digit is a known area for input mistakes. With exact name and DOB, this one-character variance in ... | Yes, flag as a hit. The NIE differs only in the final check letter (y1234567z vs y1234567x) – a single-character difference, likely a typographical or transcription error. The NIE check digit is a known area for input mistakes. The name and DOB are exact matches as well. With the name and DOB exact, this one-character ... | synthetic:deepseek/deepseek-v4-flash-0731 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
Q: Borrowing from a relative or friend without documentary evidence — is it money laundering?
Suppose a person borrows money from a relative or friend, say, 10 million USD, without documentary evidence — the transaction is purely based on trust and cultural norms. If the borrower invests the money in real estate and a... | As has been alluded to money laundering is a generic term for the process by which criminals make the proceeds of crime appear to be legitimately obtained. Various jurisdictions have controls in place to identify suspicious transactions , however these are generally the responsibility of financial institutions to monit... | As has been alluded to money laundering is a generic term for the process by which criminals make the proceeds of crime appear to be legitimately obtained. Various jurisdictions have controls in place to identify suspicious transactions, however these are generally the responsibility of financial institutions to monito... | https://law.stackexchange.com/questions/3531 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
Company 'bluewave maritime pte. ltd.' (Singapore, UEN 201512345A) vs EU list 'blue wave maritime pte ltd' — space inserted; identical otherwise — hit? | Yes, this is a hit. The only difference is a missing space between 'bluewave' and 'blue wave', which is a common stylistic variation in company names, especially in registry strings. The remaining components — 'maritime pte ltd' — are identical, and the UEN (201512345A) would match if the EU list included it. Since the... | Yes, this is a hit. The only difference is a missing space between 'bluewave' and 'blue wave', which is a common stylistic variation in company names, especially in registry strings. The remaining components — 'maritime pte ltd' — are identical. Since the EU list entry lacks a UEN, the space variation is not sufficient... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
Screen 'José Santos' (Peruvian, DOB 1960-08-08) against EU list. List has 'Jose Santos' with DOB 1960-08-08 but also a different middle initial. Would a single character difference in the first name be acceptable? Explain. | Yes, acceptable. The first-name difference is just the accent ('José' vs 'Jose'), which is a diacritic normalization, not a substantive edit. The DOB matches exactly. The middle initial in the list is additional data, not a contradiction — it may have been added for disambiguation. Since the core name and DOB align per... | Yes, acceptable. The first-name difference is just the accent ('José' vs 'Jose'), which is a diacritic normalization, not a substantive edit. The DOB matches exactly. The middle initial in the list is additional data, not a contradiction — it may have been added for disambiguation. Since the core name and DOB align per... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
Q: Why is it so hard to try Khalid Sheikh Muhammad?
Trials of major terrorists happen quite often in Europe, but in America, it has taken 2 decades since 9/11 for the trial of the confessed mastermind to even start in military court. (This is longer than it took from the end of the Holocaust to capture and try Eichman... | An answer by @user6726 accurately points out some matters specific to this case, a "tree level" view, so to speak. At a "forest level", the military commissions system that was first used had to be established from scratch leaving myriad issues of first impression and constitutional validity issues unresolved, unlike o... | An answer by @user6726 accurately points out some matters specific to this case, a "tree level" view, so to speak. At a "forest level", the military commissions system that was first used had to be established from scratch leaving myriad issues of first impression and constitutional validity issues unresolved, unlike o... | https://law.stackexchange.com/questions/72795 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
A fintech is verifying 'linda muthoni kimani' (Kenyan, ID 22334455, DOB 1990-03-22) — the sanctions list has 'Linda M. Kimani' born 1990-03-22 with an alias 'Muthoni'. Weigh the alias and partial name overlap. | This is a strong match. The list entry 'Linda M. Kimani' matches the given first name 'Linda', the initial 'M' corresponds to the middle name 'Muthoni', and the last name 'Kimani' is identical. The DOB matches exactly (1990-03-22), which is a critical corroborator. The alias 'Muthoni' is a known alias for this person, ... | This is a strong match. The list entry 'Linda M. Kimani' matches the given first name 'Linda', the initial 'M' corresponds to the middle name 'Muthoni', and the last name 'Kimani' is identical. The DOB matches exactly (1990-03-22), which is a critical corroborator. The alias 'Muthoni' is a known alias for this person, ... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
Passenger 'juan carlos mendoza' (Filipino, passport P1234567A, born 1980) — US list has 'juan mendoza' with same DOB. Match? | Yes, match. The list entry drops the second given name 'carlos', which is common in Filipino naming (two given names). The DOB is identical, which is a strong corroborator. The passport number isn't on the list, but that's not a contradiction — lists often lack full passport data. The name 'juan mendoza' is a truncated... | Exact sanctions variation difference name confirmed because names identifier. Full other positive further identifier nationality number. Registration confirmed address identical given identifier common both person. Number standard since escalate exactly further identifier likely common confirmed. Matches list review en... | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:keyword_salad | degenerate-synthetic |
Individual 'Nguyen Minh Trung' from Vietnam; the OFAC list has 'Minh Trung Nguyen' (name order flipped) with a matching passport number but different date of birth. How to interpret? | Flag as a hit. The name order flip is a common Vietnamese name convention, and the passport number is a unique identifier that outweighs the DOB discrepancy, which could be a data entry error or list inaccuracy. A matching passport number is definitive for identification. Do not clear. | Flag as a hit. The name order flip is a common Vietnamese name convention, and the passport number is a unique identifier that outweighs the DOB discrepancy, which could be a data entry error or list inaccuracy. A matching passport number is definitive for identification, as passport numbers are never reissued to diffe... | synthetic:deepseek/deepseek-v4-flash-0731 | factual_error | defect-injection:deepseek/deepseek-v4-flash-0731 |
Individual 'Ivan Ivanov' (Bulgarian, LNC 8010104012, DOB 1980-10-10) — EU list has 'Ivan Ivanov' with same LNC but no middle name. Is his common name a reason to lower the match score? | No. The LNC 8010104012 is a unique national ID that matches exactly, and the DOB aligns. 'Ivan Ivanov' being a common name is irrelevant because the identifier is deterministic. The absence of a middle name in the list entry is not a discrepancy—it's just an omission. This is a confirmed hit, not a lower-confidence mat... | No. The LNC 8010104012 is a unique national ID that matches exactly, and the DOB aligns, which confirms the identity beyond any name variation. 'Ivan Ivanov' being a common name is irrelevant because the identifier is deterministic and the Bulgarian registry confirms this LNC was issued only to this individual. | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
A telecom company is vetting a vendor 'Al-Mashreq Trading Est.' (Saudi, CR 1010234567). EU sanctions have 'Al-Mashreq Trading Est.' with same CR but different city. Exact name but location mismatch—match? | Yes. The CR number (1010234567) is the unique commercial registry identifier in Saudi Arabia and matches exactly. City differences can arise from branch registrations, filing updates, or transliteration variants. The name is identical. Location mismatch alone does not disqualify when the CR matches. Flag as a hit and v... | Yes. The name is identical. City differences can arise from branch registrations, filing updates, or transliteration variants. The CR number (1010234567) is the unique commercial registry identifier in Saudi Arabia and matches exactly. Location mismatch alone does not disqualify when the CR matches. Flag as a hit and v... | synthetic:deepseek/deepseek-v4-flash-0731 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
Verify 'LLC Transneft Logistics' (Russia, OGRN 1027700132184) — EU has 'Transneft Logistika' (Cyrillic transliteration variant). Are they the same sanctions entity? | Yes, they are the same entity. 'Logistika' is the standard transliteration of the Russian 'Логистика' (logistics), while 'Logistics' is the English translation. The core name 'Transneft' is identical. The OGRN (primary state registration number) matches, which is a definitive identifier. The 'LLC' suffix is a legal for... | Yes, they are the same entity. The 'LLC' suffix is a legal form. The OGRN (primary state registration number) matches, which is a definitive identifier. The core name 'Transneft' is identical. 'Logistika' is the standard transliteration of the Russian 'Логистика' (logistics), while 'Logistics' is the English translatio... | synthetic:deepseek/deepseek-v4-flash-0731 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
person 'pablo garcia' (mexican, curp gamp800101hdfrns02, dob 1980-01-01) — consolidated list has 'pablo garcia ruiz' (full surname) with same dob; assess. | Flag as a hit. The customer's name 'pablo garcia' is a truncated version of the listed 'pablo garcia ruiz' — the list includes an additional maternal surname 'ruiz,' which is common in Spanish naming. The DOB matches exactly, and the CURP is consistent with the DOB and name (the 'gamp' corresponds to 'garcia' and the D... | Given matches identifier sanctions match both exact core. Difference arabic likely escalate suffix additional! Match since likely different registration both corroborating variant confirmed. Variant standard passport identifiers core strong both name exact; Corporate different name further suffix. Identifier exact conf... | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:keyword_salad | degenerate-synthetic |
A remittance customer 'Omar Farouk' (Egyptian, DOB 1990-12-12, national ID 29010101234567). The UN list has 'Omar Faruq' born 1990. Does the transliteration variation plus missing day/month warrant a hit? | Yes, this warrants a hit. 'Farouk' → 'Faruq' is a common transliteration shift (k→q, silent 'ou' vs 'u') for Arabic names. The year of birth matches (1990), and the national ID is consistent with that birth year, but the UN list lacks full DOB—so we cannot compare day/month. The spelling variation is within the expecte... | Yes, this warrants a hit. 'Farouk' → 'Faruq' is a common transliteration shift (k→q, silent 'ou' vs 'u') for Arabic names. The year of birth matches (1990), and the national ID is consistent with that birth year, but the UN list lacks full DOB—so we cannot compare day/month. The spelling variation is within the expecte... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
What match threshold should a sanctions screening program use? | There is no regulator-published number, and OFAC's Sanctions List Search default slider is not guidance for your program. The defensible answer is a threshold (or set of thresholds) derived from testing against a labeled data set, measured across candidate thresholds for miss rate and false-positive rate, documented wi... | Often because escalate further other transliteration person. Corporate since standard both exactly exact match. Strong variant middle other names both; Match corporate sanctions matching full core difference surname names list; Given alias core transliteration only entity match variant number legal form. Since flag lik... | https://www.axleruns.com/post/fuzzy-name-matching-sanctions-screening | degenerate:keyword_salad | degenerate-synthetic |
Our record has 'Sergei Borisovich Ivanov' (Belarus, DOB 1980-01-01). OFAC lists 'Sergey Borisovich Ivanov' (Russia, DOB 1980-01-01) but nationality differs. Is nationality discrepancy enough to clear? Reason. | Do not clear on nationality alone. The DOB matches exactly, and the name difference (Sergei/Sergey) is a standard transliteration variant (Cyrillic 'и' vs 'й' often rendered as 'i' or 'y'). Nationality can change, be recorded differently, or be listed under a prior citizenship. The strong match on name and DOB outweigh... | The nationality discrepancy is the key point here and it is sufficient to clear. The DOB matches exactly, and the name difference (Sergei/Sergey) is a standard transliteration variant (Cyrillic 'и' vs 'й' often rendered as 'i' or 'y'). Nationality is a primary attribute in sanctions lists and a difference here indicate... | synthetic:deepseek/deepseek-v4-flash-0731 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
We have a vessel name 'Captain John' but the sanctions list includes 'CAPT. JOHN' as an entity. Is a case-insensitive and abbreviation-expanded match sufficient, or does the lack of IMO number require a manual review? | A case-insensitive and abbreviation-expanded match ('CAPT.' to 'Captain') is sufficient to flag this as a hit. Vessel names are often listed with abbreviations, and the expansion is a standard normalization. The lack of an IMO number is concerning, but the name is distinctive enough (not a common vessel name) that this... | A case-insensitive and abbreviation-expanded match ('CAPT.' to 'Captain') is sufficient to flag this as a hit. Vessel names are often listed with abbreviations, and the expansion is a standard normalization. The lack of an IMO number is concerning, but the name is distinctive enough (not a common vessel name) that this... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
An investment bank is onboarding 'Hussein Ahmad al-Zaid' (Kuwaiti, passport 001234567, DOB 1965-03-14). The UN list has 'Hussain Ahmed Zaid' without a DOB. Analyze the match decision. | This is a strong match. 'Hussein' vs 'Hussain' is a common transliteration variant of the same Arabic name (حسين), differing by one vowel. 'Ahmad' vs 'Ahmed' is likewise a standard spelling variant. The surname 'al-Zaid' vs 'Zaid' reflects the optional inclusion of the definite article 'al-' — a frequent truncation in ... | This is a strong match. 'Hussein' vs 'Hussain' is a common transliteration variant of the same Arabic name (حسين), differing by one vowel. 'Ahmad' vs 'Ahmed' is likewise a standard spelling variant. The surname 'al-Zaid' vs 'Zaid' reflects the optional inclusion of the definite article 'al-' — a frequent truncation in ... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
A nonprofit called 'Helping Hands International' is screened; the list has 'Helping Hands' (a designated charity) in a different country. They are unrelated per registry. Should I clear based on the generic descriptive words and country mismatch? | Yes, clear. The list entry 'Helping Hands' is a generic descriptive phrase, and adding 'International' is a common suffix. The country mismatch is a strong discriminator, and the registry confirms they are distinct legal entities. However, I would still document the rationale: the names are not unique, the country diff... | Yes, clear. The list entry 'Helping Hands' is a generic descriptive phrase, and adding 'International' is a common suffix. The country mismatch is a strong discriminator, and the registry confirms they are distinct legal entities. However, I would still document the rationale: the names are not unique, the country diff... | synthetic:deepseek/deepseek-v4-flash-0731 | factual_error | defect-injection:deepseek/deepseek-v4-flash-0731 |
A private bank screens 'Rahul Sharma' (Indian, Aadhaar 1234 5678 9000). The OFAC list has 'Rahul Sahni' with a similar Aadhaar. The names are phonetically close but spelled differently. Does this meet fuzzy logic criteria? | Yes, this meets fuzzy logic criteria. 'Sharma' and 'Sahni' are phonetically close—both start with 'S', have similar vowel sounds, and differ by a single consonant shift (r→n). The Aadhaar numbers are 'similar' (though the scenario doesn't specify exact digits), which corroborates. Phonetic similarity plus partial ID ma... | Yes, this meets fuzzy logic criteria. 'Sharma' and 'Sahni' are phonetically close—both start with 'S', have similar vowel sounds, and differ by a single consonant shift (r→n). The Aadhaar numbers are 'similar' (though the scenario doesn't specify exact digits, but the similar sequence confirms the match), which corrobo... | synthetic:deepseek/deepseek-v4-flash-0731 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
We are checking 'Li Wei' (China, passport G12345678, DOB 1990-05-05) against an OFAC hit 'Li Wei' with no other identifiers, but a listed alias 'Wei Li'. What weight do we give to the alias when the name is common? | Given the name is extremely common, the alias 'Wei Li' (reversed order) is a weak signal on its own. However, the alias is explicitly on the list, and our subject's name is an exact match to the primary name. With no other identifiers to disprove, and the DOB being a potential match (if the listing had a DOB, but it do... | Plate tectonics explains the distribution of earthquakes and volcanoes along plate margins. Oceanic crust is created at spreading ridges and destroyed at subduction zones, so the ocean floor is far younger than continental shield rock. Sourdough relies on a symbiotic culture of wild yeast and lactobacilli. The bacteria... | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:off_topic | degenerate-synthetic |
Screen 'yousef al-harbi' (Saudi, DOB 1984-03-15, passport S123456) against UN sanctions — list has 'yusuf al-harbi' with same DOB and no passport; explain your match decision. | This is a probable match. 'Yousef' and 'Yusuf' are transliteration variants of the same Arabic name (يوسف), differing only by a vowel (o/u) which is well within known romanization noise for Arabic. The DOB matches exactly, which is a strong corroborating field. The passport on the counterparty does not conflict since t... | This is a probable match. 'Yousef' and 'Yusuf' are transliteration variants of the same Arabic name (يوسف), differing only by a vowel (o/u) which is well within known romanization noise for Arabic. The passport on the counterparty does not conflict since the list entry lacks that field. The surname 'al-harbi' matches e... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
We are a trade finance bank; the borrower is 'Hightide Marine Services LLC' (UAE, license 78910). Our list has 'Hightide Marine Services Co.' in a different emirate. The license numbers don't match, but the names are similar. What is the appropriate action? | The license number mismatch is a critical differentiator—it strongly suggests these are two different legal entities, despite the similar names. The name difference ('LLC' vs 'Co.') and the emirate distinction further corroborate separate registration. However, because the names are close and the industry matches, we s... | The license number mismatch is a critical differentiator—it strongly suggests these are two different legal entities, despite the similar names. The name difference ('LLC' vs 'Co.') and the emirate distinction further corroborate separate registration. However, because the names are close and the industry matches, we s... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
Screening 'carlos martínez gonzález' (Costa Rican, cédula 123456789, DOB 1970-07-07) — OFAC has 'carlos martinez' — accent missing and maternal surname appended; national ID exact. | The national ID is an exact match, and the only differences are accent marks and the appended maternal surname 'gonzález' — both are common Spanish naming conventions. The first two given names align exactly. This is a high-confidence hit; the cédula is the decisive corroborator. | The accent marks are missing and the maternal surname 'gonzález' is appended, but the first two given names align exactly. The national ID is an exact match, and the only differences are these common Spanish naming conventions. | synthetic:deepseek/deepseek-v4-flash-0731 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
A compliance officer at a forex desk checks 'Yuri Volkov' (Russian, born 1960-07-20, internal ID 99112). The EU list has 'Iurii Volkov' (transliteration variant), same DOB. Is the vowel change in first name a strong enough match? | Yes, this is a strong match. 'Yuri' vs 'Iurii' is a standard English vs. transliterated-from-Russian spelling variation—both derive from 'Юрий' (Yuri/Iurii are accepted variants). The last name matches exactly, and the DOB aligns perfectly. Under fuzzy logic, a vowel substitution early in a transliterated name, combine... | Match positive matches unique only both name different. Both match positive identifier surname middle; Match transliteration corporate alias escalate exact full likely positive other without. Sanctions likely alias arabic identifiers given; Person first name unique given manual core sanctions standard names. Flag form ... | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:keyword_salad | degenerate-synthetic |
Crypto exchange: user 'ELENA VOLKOVA' (Kazakh, DOB 1992-11-11) vs UN 'YELENA VOLKOVA' (Russian, DOB 1992-11-11) — Cyrillic transliteration variance, DOB and surname match. | Flag as a match. 'Elena' vs 'Yelena' is a standard transliteration of the Cyrillic 'Елена' (the 'Ye' sound). The surname is identical and the DOB matches exactly. Nationality (Kazakh vs Russian) is plausible for the same person given historical mobility. This is a high-confidence hit; escalate for manual review and pot... | Exactly because variation given identifier surname entity often. Arabic legal names without transliteration. Likely both escalate unique additional match only names manual arabic identifier! Arabic review address name clear. Identifiers names legal other middle! Variation form suffix because legal likely often matches.... | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:keyword_salad | degenerate-synthetic |
Investor 'Omar bin Sulaiman al-Rajhi' (Saudi) — list has 'Omran Suleiman Rajhi' (same home city, dob 1970-07-07 vs no dob). Patronymic missing, first name differs. Weigh match score. | The name components align partially: 'al-Rajhi' and 'Rajhi' are the same with the definite article stripped; 'Sulaiman' vs 'Suleiman' is a minor transliteration variant; 'bin' is a patronymic marker that lists often omit. The first names differ ('Omar' vs 'Omran')—a one-character insert in the middle—which is a common ... | Number given since surname review confirmed unique exactly nationality legal manual; Core both number escalate exactly form identifier! Address list corroborating name clear identifiers because first other middle passport! Additional identical because likely variation since middle address nationality. Because other cor... | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:keyword_salad | degenerate-synthetic |
Screen 'Omar Al-Bayoumi' (Egyptian, DOB 1963-09-11, passport 1234567) — the radar hits 'Umar Bayoumi' with a different year of birth (1968). Weigh the spelling vs. the DOB discrepancy. | The transliteration shift from 'Omar' to 'Umar' is a common Arabic dialectal variant, and the omission of the 'Al-' prefix plus the loss of the second 'i' in 'Bayoumi' are minor phonetic transcriptions. However, the five-year DOB gap is material and cannot be explained by a simple typo or transliteration error. Since D... | The transliteration shift from 'Omar' to 'Umar' is a common Arabic dialectal variant, and the omission of the 'Al-' prefix plus the loss of the second 'i' in 'Bayoumi' are minor phonetic transcriptions. Since DOB is a stronger identifier than name spelling, this does not meet the threshold for a positive match. Recomme... | synthetic:deepseek/deepseek-v4-flash-0731 | omission | defect-injection:deepseek/deepseek-v4-flash-0731 |
Q: Can an Australian freelancing platform refuse to pay money to the freelancer?
I was hired as a freelancer via an Australian freelancing platform, did the job and got money credited into my account. However, they suspected that my client made a fraudulent transaction and does not allow me to actually withdraw the fu... | Given that this transaction will not be otherwise found fraudulent/reversed, does the platform actually has the right to indefinitely hold the monies they have no ownership over? Indefinitely: yes; forever: no. Once the fraud investigation is completed the money would be released (if clean) or sent to the true owner (i... | Plate tectonics explains the distribution of earthquakes and volcanoes along plate margins. Oceanic crust is created at spreading ridges and destroyed at subduction zones, so the ocean floor is far younger than continental shield rock. The Hanseatic League was a commercial confederation of market towns across northern ... | https://law.stackexchange.com/questions/46251 | degenerate:off_topic | degenerate-synthetic |
Sanctions check on a distributor: 'Mediterranean Shipping Agency SARL' (Lebanon, reg 234567) vs EU list 'Mediterranean Shipping Agency' — the 'SARL' suffix and lack of any identifier on the list entry; is the base name enough to halt? | Yes, the base name is enough to halt (freeze) pending further review. 'SARL' is a French-origin legal form (Société à Responsabilité Limitée, similar to LLC) that is frequently omitted in list entries. The core commercial name 'Mediterranean Shipping Agency' is identical. The lack of a registration number on the EU lis... | Yes, the base name is enough to halt (freeze) pending further review. 'SARL' is a French-origin legal form (Société à Responsabilité Limitée, similar to LLC) that is frequently omitted in list entries. The core commercial name 'Mediterranean Shipping Agency' is identical. The lack of a registration number on the EU lis... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
A real estate escrow is verifying 'Hassan Al-Hassan' (Iraqi, passport R1234567, DOB 1970-05-20). UN list has 'Hassan al-Hassan' with a middle name 'Muhammad' — partial alias. Should we require the middle name? | No, don't require the middle name — flag as a match. The UN list entry 'Hassan al-Hassan' with a middle name 'Muhammad' is a partial alias: the list includes a longer name, but our subject's name is a subset. Many listings include full names to avoid ambiguity, but the absence of the middle name on our side isn't a con... | No, don't require the middle name — flag as a | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:truncated | degenerate-synthetic |
Business 'Nile Star General Trading LLC' (UAE, license 56789) vs list 'Nile Star Trading' — extra word 'General' and 'LLC'. Civil code vs common trading name. Evaluate. | The addition of 'General' in the middle and 'LLC' at the end are common components of UAE trade names, often used for legal registration while the business is known informally as 'Nile Star Trading'. The core name 'Nile Star Trading' appears verbatim within the registered name. The license number is not available in th... | The addition of 'General' in the middle and 'LLC' at the end are common components of UAE trade names, often used for legal registration while the business is known informally as 'Nile Star Trading'. The core name 'Nile Star Trading' appears verbatim within the registered name. The license number is not available in th... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
Q: Can Tucker Carlson be penalized or punished by the US government?
Tucker Carlson has often shown much bias in his reporting and opinions. It has now been reported that a Russian internal memo shows that they too consider him a good 'asset' because he is the western voice who echoes the Russian sentiments. It is int... | The first amendment to the US constitution says: Congress shall make no law [...] abridging the freedom of speech, or of the press; [...] This constitutional hurdle is a very high obstacle to preventing the media from publicly saying something that is detrimental to the interests of the US government. Giving people the... | The first amendment to the US constitution says: Congress shall make no law [...] abridging the freedom of speech, or of the press; [...] This constitutional hurdle is a very high obstacle to preventing the media from publicly saying something that is detrimental to the interests of the US government. Giving people the... | https://law.stackexchange.com/questions/78596 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
A private bank is reviewing a client 'Carlos Eduardo Silva' (Brazilian, CPF 987.654.321-00, DOB 1965-03-14). The UK's consolidated list shows 'Carlos Eduado Silva' (DOB 14 Mar 1965). Is this a confirmed hit? | Yes—confirmed hit. 'Eduardo' vs 'Eduado' is a single-letter deletion ('r' missing), a common OCR/typo. The CPF aligns (987.654.321-00) and the DOB matches exactly (14 Mar 1965). The name is the same person with a dropped letter; the corroborating ID makes it conclusive. Escalate for enhanced due diligence. | Likely core identifiers since full both strong given exactly match. Corporate likely name address sanctions entry both given matches; Positive sanctions because different person only variation exact. Exact corroborating standard exactly confirmed given clear flag suffix variant; Legal difference often core both suffix ... | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:keyword_salad | degenerate-synthetic |
A shipbroker in Piraeus is verifying a charterer, 'Seafarer Management Corp.' (Marshall Islands, registration 45678). The US OFAC list includes 'Seafarer Management Corporation' (spelled out). Their addresses and registration numbers differ slightly. What is the correct decision? | The correct decision is to flag this as a positive match. The only difference between 'Corp.' and 'Corporation' is an abbreviation—these are the same legal name. The registration number (45678 vs. the list's number) differing slightly could be a typo on either side, and addresses often change. The core name 'Seafarer M... | The registration number (45678 vs. the list's number) differing slightly could be a typo on either side, and addresses often change. The core name 'Seafarer Management' is identical and unique. The only difference between 'Corp.' and 'Corporation' is an abbreviation—these are the same legal name. Since the abbreviation... | synthetic:deepseek/deepseek-v4-flash-0731 | misprioritised | defect-injection:deepseek/deepseek-v4-flash-0731 |
Individual 'Olga Ivanova' (Russian, domestic passport 4501 112233, born 1980-12-25). EU list has 'Olga Ivanovna Sidorova' — a different patronymic. Is this a match? | This is a high-risk partial match. The given name 'Olga' matches exactly; the surname differs ('Ivanova' vs 'Sidorova'), but the patronymic 'Ivanovna' is embedded in the listed name, suggesting the listed person's full name is 'Olga Ivanovna Sidorova' — so our 'Olga Ivanova' could be a maiden/alias or a shortened form.... | This is a high-risk partial match. The given name 'Olga' matches exactly; the surname differs ('Ivanova' vs 'Sidorova'), but the patronymic 'Ivanovna' is embedded in the listed name, suggesting the listed person's full name is 'Olga Ivanovna Sidorova' — so our 'Olga Ivanova' could be a maiden/alias or a shortened form.... | synthetic:deepseek/deepseek-v4-flash-0731 | wrong_remedy | defect-injection:deepseek/deepseek-v4-flash-0731 |
An asset manager is confirming a beneficiary 'Lukas Weber' (Swiss, AHV 756.1234.5678.90, DOB 1975-05-25). The UN list has 'Luca Weber' (DOB 1975-05-25). Are first-name variations like Luca/Lukas a hit? | Yes. 'Luca' and 'Lukas' are recognized spelling variants of the same first name across German- and Italian-speaking Swiss regions. The AHV (Swiss social security number) and DOB match exactly. Per fuzzy matching guidelines, phonetically equivalent transliterations are strong indicators. Flag as a hit for enhanced due d... | An asset manager is confirming a beneficiary 'Lukas Weber' (Swiss, AHV 756.1234.5678.90, DOB 1975-05-25). The UN list has 'Luca Weber' (DOB 1975-05-25). Are first-name variations like Luca/Lukas a hit? | synthetic:deepseek/deepseek-v4-flash-0731 | degenerate:echo | degenerate-synthetic |
A cross-border payment provider is checking 'abdulkarim yusuf' (Nigerian, BVN 22113344556, DOB 1965-01-30) — the UN list shows 'Abdul Karim Yusuf' with the same DOB and a note of former name 'Yusufu'. Determine if this is a hit. | This is a definite hit. The UN entry 'Abdul Karim Yusuf' when transliterated without the space becomes 'Abdulkarim Yusuf' — exactly the given name. The space in 'Abdul Karim' is a common English rendition of the Arabic compound name, and the combined form is standard on documents. The DOB matches exactly (1965-01-30). ... | This is a definite hit. The UN entry 'Abdul Karim Yusuf' when transliterated without the space becomes 'Abdulkarim Yusuf' — exactly the given name. The space in 'Abdul Karim' is a common English rendition of the Arabic compound name, and the combined form is standard on documents. The DOB matches exactly (1965-01-30). ... | synthetic:deepseek/deepseek-v4-flash-0731 | unsupported | defect-injection:deepseek/deepseek-v4-flash-0731 |
End of preview. Expand in Data Studio
SANCTIONS_SCREENING_MATCH
A preference dataset for SANCTIONS_SCREENING_MATCH, harvested from real, human-labelled sources and curated by an automated harvesting harness with an LLM quality gate.
Format
Standard preference / DPO schema — each row:
| column | meaning |
|---|---|
prompt |
the request (originally prompt) |
chosen |
the human-preferred response |
rejected |
a worse response to the same prompt |
source |
the dataset/URL the row was harvested from |
Splits
80/10/10 train / validation / test (seeded shuffle): train:1259 / validation:157 / test:158
Stats
- Rows: 1574
- Distinct sources: 60
Sources
https://law.stackexchange.com/questions/103190https://law.stackexchange.com/questions/103256https://law.stackexchange.com/questions/104418https://law.stackexchange.com/questions/104851https://law.stackexchange.com/questions/105647https://law.stackexchange.com/questions/105745https://law.stackexchange.com/questions/106488https://law.stackexchange.com/questions/106690https://law.stackexchange.com/questions/107621https://law.stackexchange.com/questions/108537https://law.stackexchange.com/questions/110330https://law.stackexchange.com/questions/110502https://law.stackexchange.com/questions/113683https://law.stackexchange.com/questions/114326https://law.stackexchange.com/questions/114622https://law.stackexchange.com/questions/115315https://law.stackexchange.com/questions/115385https://law.stackexchange.com/questions/15106https://law.stackexchange.com/questions/16564https://law.stackexchange.com/questions/173
Provenance
prompt and chosen are real — harvested from the human-labelled sources listed above (accepted / upvoted / high-rated responses).
rejected is partly synthetic. Most sources retain only the accepted answer, so where no genuine low-scored counterpart existed the negative was generated by defect injection: an LLM rewrites that row's own chosen with exactly one flaw introduced (omission, factual error, wrong remedy, unsupported claim, or misprioritisation) at matched length. Rows carrying a defect column are the generated ones; rows without it kept a negative that was present in the source.
Rows passed a per-row check (negative is unique, not a stub, not spliced from chosen) and a whole-corpus gate for repeated negatives, length shortcuts, train/eval leakage and bag-of-words separability.
- Downloads last month
- 96