2. Watchlist Screening
• Что это такое:
– Бизнес-процесс выявления потенциального совпадения между записью
клиента или любой другой (клиент, поставщик, бизнес-партнёр, сотрудник)
которая может появиться в публичном или частном ‘списке’
– Возможность избежать ведения бизнеса с нежелательными физическими или
юридическими лицами
• Почему это важно:
– Международные и региональные законы
• Борьба с отмыванием денег
• Предотвращение финансирования терроризма
– Определение потенциального отмывания денег, взяточничества, коррупции и/или финансирования терроризма
• Необходимость в решении следующего поколения:
– Улучшенные результаты
•
•
•
•
Лучше выявление соответствий
Снижение ложных срабатываний
Снижение операционных издержек
Лучше масштабируемость
“Know Your Customer” (KYC) and
“Know Your Customer” (KYC) and
Enhanced Due Diligence (EDD)
Enhanced Due Diligence (EDD)
obligations
obligations
•• Bank Secrecy Act 1970
Bank Secrecy Act 1970
•• Foreign Corrupt Practices Act 1977
Foreign Corrupt Practices Act 1977
•• Money Laundering Control Act of 1986
Money Laundering Control Act of 1986
•• Terrorism Act 2000
Terrorism Act 2000
•• Financial Services & Markets Act 2000
Financial Services & Markets Act 2000
•• USA Patriot Act 2001
USA Patriot Act 2001
•• Proceeds of Crime Act 2002
Proceeds of Crime Act 2002
•• EU 3rd Money Laundering Directive 2007
EU 3rd Money Laundering Directive 2007
•• Money Laundering Regulations 2007
Money Laundering Regulations 2007
•• UK Bribery Act 2010
UK Bribery Act 2010
•• ...
...
2
3. Данные имеют сложную структуру
Примерно
Примерно
300,000 имён
300,000 имён
только в
только в
списке US
списке US
OFAC...
OFAC...
3
4. Данные имеют сложную структуру
19 Aliases and 4 Locations
Простого матчинга
недостаточно
10 Aliases and 2 Date of Births
5. Информация о клиентах имеет сложную структуру
Типичные примеры:
1. ‘Overfilling’ of name data
• Списки и клиентские данные
имеют часто включают
непоследовательную,
неточную и неполную
информацию
• Screening against data that is
not fit for purpose reduces
screening accuracy,
increasing risks and costs
2. Poor spelling of name and address information
3. Multiple names stored in a single field
4. Name information ‘misfielded’ into addresses
5. Date of Birth information in various formats
6. Entities and individuals mixed together
7. Non-standard name constructs
8. Poorly fielded address information
9. Non-standard country information
6. Ключевые требования к Watchlist Screening
• Оптимизация источников данных (списков с
проверочными данными, списков клиентов и пр.)
– Любой список из любого источника
– Понимание, стандартизация и обогащение
• Лучшие в своём классе технологии матчинга
–
–
–
–
–
Эквиваленты имён и альяся
Множество языков и скрипты
Эффективный масщтабируемый матчинг
Открытый настраиваемый механизм матчинга
Приоритезация результатов в соответствии с рисками
• Управление результатами
– Эффективный интегрированное ручное управление
– Конфигурирование автоматических правил в соответствии с
рисками
– Показатели процесса и управление
– Аудит
6
7. Приложение Watchlist Screening
Prepare &
Optimize
Case
Management
Understand
500+ match
rules
Prioritized alerts
Structure
Aliases/name
equivalents
Customizable
workflow
Standardize
Watch lists
•Public lists (HMT,
OFAC, EU, UN, etc.)
•Commercial Lists
•Private lists
Match &
Score
Translation/
transliteration
Reporting &
metrics
Enrich
Match & risk
scoring
Audit trail
Customer lists
•Multiple systems
•Inconsistent
information &
standards
7
8. Более 500 правил поиска дубликатов
С лёгкой конфигурацией
Матчинг
использует
“бизнесправила” с
непонятными
комментариями
8
9. Работа с международными данными
• Спецсписки публикуются на
латинском языке, однако клиентские
данные могут быть на любом языке...
• Требования к сканированию:
– Транслитерация
– Транскрипция
– Библиотеки имён в альтернативных
формах, прозвища и пр.
• Свойства:
– Межязыковое выявление
– 45 языков и 57 стран
•
•
•
•
•
•
Cyrillic
Greek
Chinese
Arabic
Hangul (Korean)
Kanji, Hiragana and Katakana
(Japanese)
– Расширенные библиотеки имён
•
•
•
•
Chinese
Korean
Arabic
Japanese
9
10. Эквиваленты имён
• Транслитерация –
преобразования
символов; может теряться
фонетическая
информация
• Транскрипция –
преобразование
фонетического
представления
• Перевод –
словарные
эквиваленты,
включая прозвища
Editor's Notes
What do watch lists include?
Watch lists published by government agencies (i.e. HMT in UK) include lists of sanctioned individuals, entities, de-barred countries, terrorist organisations, etc. Doing business with any individual or organisation listed on a government watchlist risks prosecution in the form of serious fines and potentially imprisonment
Commercial watch lists published by the likes of Dow Jones, World-Check and others include entries included on government sanction lists as well as what we call “Politically Exposed Persons” or PEPs. Keeping it simple, a PEP can be thought of as a high ranking government or public official and as such, may be at greater risk of being exposed to bribery & corruption. Legislation therefore stipulates that organisations must apply enhanced due diligence when doing business with PEPs to assess the potential risk of facilitating transactions that may represent the proceeds of crime
The challenge in watchlist screening is to be able to accurately identify actual sources of risk, whilst minimising the number of alerts presented for review that turn out to be erroneous matches, otherwise known as false positives. This is where Oracle’s Watchlist screening solution differentiates itself
Here we have the same individual “Abu Bakr” with 10 aliases and showing a DoB of 20th and 10th August 1972
Is this the same person?
Run through each of the examples here - this is self explanatory
Its the combination of accurate data preparation and effective screening that makes our product unique
Firstly, we can undertake screening both in batch mode and in real-time. We also provide our own real-time UI fully integrated into case management that removes the need for our customers to create their own UI or undertake complex integration projects with customer on boarding systems
We also screen against both open source and commercial watch list providers as shown here
Then we preparing and optimize data using the data management processes such as “understand”, “structure”, “standardize” and “enrich” as shown – this is the first part of the screening process to ensure customer data is fit for the purpose of screening
Then we move into the screening stage shown as “match & score”. Here we are using techniques to review and address variances in watch list and customer data such as different formats used for data of birth, plus use of non standard country codes and name variances. At this stage, we are also normalising any name variances and converting non-Latin names into Latin format ahead of the screening process. Once we’ve done that, we prioritise Alerts using either internally derived risk scores or those available from external providers based on the secondary identifiers included by the watch list providers.
Having completed the screening process Alerts are presented to case management for review and resolution, with each type of Alert (whether PEPs, sanctions, prohibitions) being prioritized and having its own tailored workflow to align with the organizations screening policies. Finally comprehensive real-time reports and audit trails are provided to view the current status and progress of any Alerts and facilitate sharing of information with regulatory authorities if necessary
Now we can see the match rules that we have created to determine when and where we undertake a review of a potential match. These are all configurable by the customer and presented using a simple business narrative.
Over 450 match rules can be created using OWS, delivering a high level of screening sophistication..particularly important for minimising the number of matches that turn out to be false, whilst reducing the chances of a true match not being detected
You can see that we first look for where there is a match across multiple attributes as this is likely to give us a higher degree of confidence that the customer and watchlist records are for the same individual or entity and we prioritise this according within case management for review. You can see that anything with an exact name match is given a priority of 1 and where there is a match against first name or surname, this is given as priority 2.
Those with a lower level of confidence matches are given a lower priority score and will appear lower down the list of alerts for review within the case management view.
Designed for organisations seeking a global screening solution
Uses transliteration and transcription techniques
Where client data is held in multiple writing systems – Customers, Suppliers, Partners, Employees
Simplifies the process of converting non-Latin client data into Latin form
As used by the majority of Sanctions & Watch List Providers
Enhances the data optimisation process
Leading to more accurate identification of risk
And minimising the number of false positives
Shortens the time taken to onboard new customers in growth markets
5 different writing systems
Further 40 more in the pipeline
Multiple name equivalencies
Central, Eastern, South East Asia as well as the Middle East
Here we have an example of using transliteration techniques (character by character conversion) to convert from the Greek native writing system to the Latin equivalent