Entity Resolution:
Don't Rip & Replace
Upgrade In Place
Every organization is using one form of data matching or another, here and there, in fact almost everywhere. Whether the existing Entity Resolution (ER) is serving your KYC, AML, MDM, CRM, or fraud detection needs, almost certainly there is more Entity Resolution work to be done.
Someone, somewhere, in your organization is collecting a to-do list of Entity Resolution feature requests, accuracy tweaks, performance enhancements, bug fixes, etc. Such lists often represent years of effort and multi-millions of dollars of work. Maybe another entire re-write is what is being contemplated.
There is another option: Augment your in-house record matching with Senzing.
Because Senzing is a purpose-built AI for Entity Resolution, delivered as an SDK, developers can quickly bring Senzing online – in a production capacity – to AUGMENT YOUR EXISTING ENTITY RESOLUTION investments.
Attaching Senzing, like a sidecar, to existing record matching processes allows organizations to fast-track their business goals as they instantly retire backlogged engineering requests. This approach allows organizations to modernize and optimize while leveraging existing investments.
There are two general ways Senzing can be used to augment:
- Category 1: Senzing can be used playing a “sidecar” role – providing the existing Entity Resolution technology with additional, in-line, recommendations and insights.
- Category 2: Senzing can be used to deliver whole new capabilities – deployed more like a parallel-track.
Let’s dig in…
Category 1: Augment Existing Entity Resolution with Recommendations and Insights
In this category, Senzing can be used as a “sidecar” role, providing existing Entity Resolution algorithms with suggestions about specific matches. Depending on the kind of suggestion and overall context, suggestions may be followed no questions asked, ignored, or subject to a 3rd party review. In this implementation is augmenting to do things like combine more data sources faster or cover more edge cases.
A: Quickly combine new data sources for better analytics and decision-making
Use Senzing to create more comprehensive 360-degree views by fast tracking the on-boarding of additional data sets.
Why: The better the context, the better the insights. You may have already identified a handful or even dozens of additional data sources to be included in your machine learning models, data factory, or workflows. In supply chain projects, there may be goals to load corporate hierarchies and ultimate beneficial owners. In fraud projects, you may have an urgent need to combine historical fraud investigative data with existing KYC processes. Adding such data sources into existing Entity Resolution systems is often not so easy – resulting in an engineering backlog that can be hard to complete and retire.
How: Using Senzing’s purpose-built AI, new data sources can be quickly combined in days and weeks, not months. The Senzing breakthrough is called Principle Based Entity Resolution. This new class of matching technology allows organizations to combine diverse data sources without any training or tuning. After mapping new data to Senzing JSON, “one-click” Entity Resolution will routinely outperform humans. Further, we’ve accelerated the mapping process for many popular data sources via both free, pre-formatted, snapshots (called CORDs – Collections Of Relatable Data) as well as an ever growing list of opensource mappers.
Combine another new data source per week!
B: Strengthen existing match capabilities
Quickly cover edge case matching scenarios to drive up your matching accuracy.
Why: Senzing can be used to enhance the accuracy of your existing Entity Resolution in areas like multicultural name matching, enhanced messy address comparison, inclusion of more matching (e.g., social media handles and latitude/longitude), support for arrays of values (e.g., a list of historical addresses and multiple dates of birth). Senzing also can be used to surface difficult to detect over-matches (false positives) using its specialty “ambiguous conditions” logic. Further, Senzing’s Entity-Centric Learning algorithms can be used to surface under-matches (false negatives) – essential real-time learning needed when the use case involves clever bad actors who intentionally obfuscate their identity data.
How: Senzing provides coverage for various edge cases we’ve faced over decades of work in the field including multicultural name libraries, global address parsers trained on 1B+ records, and technique breakthroughs like Principle Based Entity Resolution. Collectively, these techniques allow Senzing to quickly provide existing Entity Resolution algorithms with recommendations to augment current matching.
Category 2: Augment Existing Entity Resolution with In-Parallel Capabilities
In this category, Senzing provides a parallel track of entity resolution. The outputs of Senzing likely being consumed by independent workflows or combined with results coming off the back end of the existing Entity Resolution processes. In this implementation, Senzing is being used to add support such things as Generative AI (Gen AI), relationship awareness, identity obfuscation for fraud detection, real-time workloads, new entity types, or support for new geographies, cultures, and languages.
A. Accelerate your GenAI initiatives
Use Senzing to keep your product on the cutting edge of the AI revolution.
Why: Generative AI (GenAI) benefits from the use of current-to-the-second enterprise data and reference data to deliver information about prospects, customers, suppliers, subjects of interest, etc. Entity resolved data delivers an essential ingredient into agentic or analytics workflows to provide current, consistent, and explainable “facts” that improves accuracy while reducing hallucinations.
How: Senzing’s purpose-built AI delivers real-time, consistent and explainable context perfectly suited for your agentic workflows, Retrieval Augmented Generation (RAG), and multi-modal reasoning architectures. As an SDK, Senzing’s composable technology adds this additional value with unprecedented simplicity i.e., no moving parts.
Example of Senzing in a Retrieval Augmented Generation (RAG) architecture.
B: Add relationship awareness for better analytics and decision-making
Use Senzing to augment in-house Entity Resolution with both disclosed and derived relationships to improve downstream analytic workflows and outcomes.
Why: Whether your product is a risk, fraud, marketing, or a pure data product, understanding how entities relate to one another improves ML/AI training models, downstream analytics, decision-making and visualization. In fact, entity resolved knowledge graphs are essential for accurate graph analytics e.g., nearest neighbor, centrality, link prediction and other popular algorithms. In risk use cases, relationship awareness improves risk detection models. In fraud use cases, this reveals hidden connections. In marketing, this enables householding and network value calculations. In supply chain, this enables supply chain visibility.
How: Senzing’s purpose-built AI, supports disclosed relationships (i.e., provided in the data) e.g., corporate hierarchies, officers, directors. Senzing also detects derived relationships e.g., entities that share an address. Out-of-the-box Senzing will augment your in-house Entity Resolution with both disclosed and derived relationships – delivered as entity-resolved graphs – for easy downstream consumption.
C: Cover advanced fraud detection use cases
Leverage the hidden connections Senzing is designed to reveal to deliver world-class fraud and risk products.
Why: Clever bad actors DO NOT use the same name, address, and ID on every record. Instead, they obfuscate their identity to circumvent data matching technologies. Most entity resolution technologies use simplistic record-to-record matching which is not able to find identity obfuscation. If your organization faces clever bad actors, Senzing’s technology is designed to find these hidden connections in your data.
How: The Senzing team has been perfecting a technique called Entity-Centric Learning™ (ECL). Using ECL, Senzing accumulates over time every variation of every match worthy feature (all names, addresses, national IDs, dates of birth, etc.). When new records are presented, Senzing is not trying to find a record – rather, Senzing is trying to find a matching entity. This distinction is so essential, without ECL existing bad guy hunting systems only find the idiots.
Nevada Gaming Control Board Excluded Person
D: Add real-time workloads to existing batch offerings
Attach Senzing to your in-house Entity Resolution to instantly support real-time use cases like customer onboarding, transaction monitoring, and other high value workloads.
Why: Entity Resolution that evaluates new data in real time – while transactions are happening – is ideal for processes like customer onboarding, continuous KYC/AML, transaction monitoring, and call center support where the information must be current to the second. Because batch-based Entity Resolution does not grow up to do real-time workloads, real-time requirements require different architectures i.e., ground-up code re-writes.
How: Senzing’s real-time AI, our 6th generation engine, is the most proven, scalable, Entity Resolution engine on the planet. The Senzing technology can support billions of records and sustain thousands of Entity Resolution events per second (load or search).
Real-time transaction support keeps the persistent entity graph current to the second.
How: Senzing’s real-time AI, our 6th generation engine, is the most proven, scalable, Entity Resolution engine on the planet. The Senzing technology can support billions of records and sustain thousands of Entity Resolution events per second (load or search).
Learn More:
E: Expand Domain Support from People to Organizations or Organizations to People
Use Senzing to quickly cover a missing entity type to support business goals related to new product offerings or competitive coverage.
Why: Whether your product is a risk, fraud, marketing, or pure data product, adding support for new entities may become a must have to drive growth or remain competitive. Pivoting from company Entity Resolution to people Entity Resolution or vice versa can be a huge undertaking making for a significant chunk of engineering backlog. This is because building Entity Resolution on business data – tailored to such features as business names, physical addresses, shipping addresses, and business licensing data has taken years of work. Extending Entity Resolution to people data with features such as name (including nicknames and maiden names), historical addresses, dates of birth, differing IDs, etc. can require significant work if this was not originally designed in.
How: Using Senzing’s purpose-built AI, people and company data are supported equally, as well as some other popular entity types like vessels and cars. Out-of-the-box Senzing can provide additional entity type coverage that can be attached to your existing internal Entity Resolution workflows.
Just as easy to do one, the other, or both at the same time.
Learn More:
F: Cover new geographies, their culture, and language scripts
Use Senzing to quickly launch coverage or product offerings into new geographies that are not innately supported by the existing Entity Resolution system.
Why: Entity resolution requires localization-specific algorithms to properly handle Roman character data versus other languages like Mandarin, Cyrillic, and Arabic. Further, Asian address parsing and comparison requires unique handling that is not well supported by algorithms built for, say, US or Canadian addresses.
How: Senzing comes out of the box with pre-built libraries and logic to support most of the world’s languages including regionalized representation nuances. Senzing culturally-aware name matching has been trained on 850M names. Senzing global address parsing and comparison even includes coverage for Mandarin to Roman character company names and addresses, itself represents years of investment. Leveraging these past and present investment areas, means users of Senzing can fast track their internal efforts by simply augmenting their in-house Entity Resolution.
Learn More:
Cross-script compare
How About This?
Conclusion
You have invested a lot already in your existing data matching capability. You likely also have a decent list of additional capabilities your business needs. Senzing may be able to fast track some of your engineering backlog.
Take Senzing for a test drive. It might be the perfect companion to accelerate your high-priority business requirements.
It’s free and easy to try. Explore the options here.