The ability to identify more than one observations that refers to the same entity, be it a person, a company or other, and being able to recognize the entity is search despite different data-structure or UI appearance. And conversely, being able to recognize when it is not the same Entity, even though it might appear to be with similar data content.
The ability to gather a 360 view of an Entity, within our databases and across other data-sources that are available to us, in a way that support the creation of Unified-view, and data-enrichment of the Entity in search (“the Golden Data Record”), in a way that enable us to take optimal decision taking, provide service offering, and enhanced communications with the Entity.
When performing Entity Resolution, our target is to identify all the data entries that are linked or referred to an Entity across multiple data sources as well as within the same database. The main requirements are:
Confidently identify your customers across omnichannel user experiences by creating unified customer profiles.
Ensure regulatory compliance with identity resolution software for GDPR compliance and accurate matching against watchlists (OFAC, AML, KYC, etc.).
Uncover deep, hidden connections in your data and detect fraudulent entities. For instance, multiple claims from the same address.
Create unified reports on all accounts related to the same beneficiary, be it saving, investing, family and business accounts, matching TIN and beneficiary accounts.
Match patient’s data across different data sources, birth certificate registries, clinical data history, different clinics and healthcare systems; allowing better and safer treatment whilst not compromising privacy.
As the world’s data is ever growing faster and faster, and whilst organizations are investing many efforts in the maintenance of the data integrity, at the end of the day our uttermost importance is on our ability TO SEE through this Ocean of Data, being able to make sense of it fast, concisely and accurately.
We want to make sure we serve well our stake-holder and most of all our customers.
This has become more and more challenging as we need to glide through different data sources, cooperation’s, exchange between different data systems and organization’s subsidiaries. And whilst we do all that – we need to maintain a Unified View on our customers or on any other Entity of interest.
In most common ER systems, there is an attempt to gather data-pieces from different sources, using rule-based methodology combined with AI, looking at various patterns and common denominators . The problem in doing that, is the lack of transparency, consistency and traceability of the process. Moreover, most technologies in use are not equipped with the ability to deal in different languages, transliterations and spelling variations. This is very important as many of the ‘Entity identifiers’ are based on texts and names.
The Phonemix methodology is very different.
The thought consideration is that ‘Entity’ is defined by Name, and therefore before anything else it will cluster data entries by Name, using the specialized ‘Phonetic Fingerprint’ technology, followed by the Phonemix-Formulae that uses advanced mathematical algorithms to refine the results using the other data fields of information. Added to that, is the ability to use the ‘Phonetic Fingerprint’ on other name fields, such Address and Email.
Finally, the system will provide a Score [%] and a Recommendation for EDD when required.
Effective resolution – is the ability to decide the importance and relevance on each data-field, whilst Normalizing the final score for assumed record-match.
Email can be a strong identifier, but how do you deal with 4 different emails for the same person.
Birthdate combined with Address and Phone can provide high probability, but what if the birthdate is 1.1.0000 ? Or the phone is 555-5555? What if the email address is invalid or empty?
NAME MATCH is Key for successful resolution!
takes all that into account, and cancel out irrelevant data, check the data with the “Phonetic Fingerprint”, that is a supreme technology for Accurate Entity-Name-Match, and then performs Fine-Tuning based on the additional information, such as Phone, Address, Email, etc.
The “Phonetic Fingerprint” identifies Names, Addresses, Emails, etc. across different languages with great accuracy, assuring no mishits whilst reducing false positives to the minimum.
Utilizing Distance Algorithms in-language and in-between languages, the Fincom.co technology can handle different transliterations, misspelling, and other spelling variations.
International companies as well as regulatory and governmental authorities are required to trace and unify information from different sources in different language origins.
Pan-European organizations must be able and handle all European languages.
An organization in the UK that has offshore offices in the UAE must be able to handle Arabic alongside English.
An American company that has subsidiary in Asia, must be able to handle Chinese or Japanese or other pictogram-based languages.
It is not very rare to find international companies that need to handle languages that are based on Latin, Cyrillic, Pictogram-based lang and Arabic, whereas the last can have over 14 different common dialects that will result in very different transliterations that are lost in legacy methodologies or rule-based methodologies.
Name Match is Key for Entity Resolution.
Phonemix Formulae – Optimize and fine tune Result’s Score
Fincom.co developed a robust sophisticated proprietary technology platform that enables to combine databases and match their entries, using over 48 algorithms even cross language(s) and/or having transliterations spelling mistakes.
FinCom’s “Phonetic Fingerprint” patented technology, create a new way for matching effectively and efficiently different data entries, and thus support a variety of advanced applications for Entity Resolution, Entity Intelligence, Data Mapping and Data unification, for a wide scope of use case and applications:
Advanced Automatic Technology for Mapping, Unifying and Navigating across Multiple & Multi-dimensional Databases