New Matching Algorithm
Our new Search API is power by a brand new screening solution. In our testing we have seen a reduction of false positives on average around 50%. This means times needed for reviewing mathces is reduced by half!
Some of the new features supported by our new algorithm are:
Person Names typology
- Structure and order detection (first name/given name, last name/surname, middle name(s), particles, suffixes…and all combinations)
- Abbreviations (J. Smith, J.L. Picard…)
- Extensive titles library (Sir, Mr, Dr, Sheik, …)
- Common aliases (Robert/Bob, Edward/Ted, ….)
- Patronyms (Vladimir Vladimirovich Putin, Abd Al Rahman ibn Muhammad, Hisham bin Hasan…)
- Teknonyms (Abu Zayn)
Entity Names typology
- Noise word elimination (OF, THE, …) in multiple languages
- Common word identification with reduced weight (BANK, INTERNATIONAL, COMPANY…)
- Legal form identification in all countries and languages (INC, LTD, OOO, SPRL, JSC…)
- Abbreviations (Aviadvigatel OJSC vs. Open Joint Stock Company Aviadvigatel)
- Synonyms / Similar words (Bureau vs Agency)
- Numbers and geographic places identification (Almaz Scientific and Production Centre, Bureau 39…)
Errors and Misspellings
- Inverted letters (John vs. Jhon)
- Doubled letters or missing doubled letters (Matter vs Mater)
- Missing or Unexpected letters (Entertainment vs Entertainement)
- Letters with similar pronunciation (Vladimirovich vs Vladimirovych) in different cultural affinities (Muhammad vs Mohammed)
- Split / Merged words (Manpower Services vs Man Power Services)
Supported Alphabets
The number of supported alphabets is growing with each new release. Currently, transliteration and automatic rescan of names is supported in the following alphabets:
- Cyrillic
- Chinese
- Japanese
- Arabic
- Greek
New Ways to Search
This new version of API introduces new ways to perform searches:
- When searching for Individuals by name, it is no longer necessary to split the name into firstname, middlename and lastname fields; You can perform a search by providing a single string for the full name. For example: “Donald John Trump”. If you know the lastname explicitly you can search like this: “Trump, Donald John”
- When you search for an Individual and provide a country, we will search against the address as well as nationality fields on our profiles; previously we only searched the addresses.
- It is now possible to include the gender when performing a search for an Individual.
- There are new controls that allow you to fine tune the matching behaviour when including the date of birth in an Individual search.
New Data Model
The data model has been redesigned to provide more granular, more consistent and more concise data points that you can use during integration. Some of the enhancements are:
- Multiple dates of birth.
- Multiple dates of death.
- Multiple nationalities.
- Alias types (including indicating the name in local script).
- External identifiers (such as VAT Number, Company Registration ID, Passport Number etc).
- Address types.
- Advanced Sanctions model.
- New Profile of Interest dataset.
- Law Enforcement and Financial Regulator lists are combined into a new Dataset called Regulatory Enforcement Lists (REL).
- Adverse Media has been renamed to Reputational Risk Exposure (RRE) and has additional datapoints now.
- Structured relationships between profiles.
- Enhanced Politically Exposed Person (PEP) data model.
- Enhanced evidence model.
Faster Response Times
We are very pleased to announce that the response times of our new Search API is over 8 times faster than our previous API. Not only is it faster but it also scales way better than its predecessor which means no slowing down at busy times of the day. Before: ~2.6 seconds, now ~0.3 seconds