Solving the puzzle of entity resolution with an easy-to-use API


Entity resolution — the process of determining when data records are about the same person, organization or other entity, despite differences in how they are described — is an important problem for companies to solve if they want to improve data quality and outcomes, but the process can be a complicated one.

Brian Macy, director of product development and operations at Senzing, said “data in every organization, is like a giant puzzle. Entity resolution helps companies figure out which pieces of information belong to which entity, and can also show how entities are related, we talk about finding who is who and who is related to whom,” he explained. 

If you don’t understand who your customers are or identify potential bad actors, you aren’t able to make decisions you can count on,” Macy explained. 

To help organizations solve this problem, Senzing provides an API that makes it easy for developers to add entity resolution capabilities to applications and services with just a few lines of code. With the Senzing API, advanced data matching and relationship discovery can be added much like you would payment processing capabilities from Stripe or communications software from Twilio. 

Senzing team members include some of the leading experts on entity resolution. The company’s founder Jeff Jonas and many members of the technical team have been working in the field for decades. Their combined experience is somewhere between 300 and 400 person years.

The Senzing entity resolution engine has AI built in that makes it smart on day one. The software also has what Macy calls “entity-centric learning” that allows it to perform highly  accurate record matching and get smarter over time as new data is added. 

Many organizations attempt to build entity resolution capabilities in house, which is often an expensive and lengthy process that fails, according to Macy. “The whole idea behind Senzing was to deliver an API that allows developers to add world-class entity resolution to their project in a couple of sprints.”

In addition to entity-centric learning, Senzing includes many other innovations in its software. One example is principle-based matching which allows Senzing entity resolution to achieve highly accurate results while eliminating the need for users to write many specific rules. It also avoids the training and tuning required by traditional probabilistic and machine learning approaches. The general set of matching principles, created based on real-world experience, saves users large amounts of time when deploying new systems and new data sources. 

The idea behind principles is as follows: If your child throws a rock at a car, and you say, “don’t throw rocks at cars” and then tomorrow they throw a baseball at a truck and you say, “don’t throw baseballs at trucks” and so on. Instead of creating individual rules, an example of a principle would be “don’t throw things at other people’s stuff!”  

The Senzing API also makes it easy to add new data sources, so organizations can start with a few sources and quickly add more over time. With homegrown and other approaches, it can take weeks or months to add new sources and tune or train the system to use them. 

Macy concluded that “entity-centric learning and principle-based [resolution] are really key to making a super easy-to-use technology for entity resolution.” It’s easy for developers to get started for free and see some results in a few minutes. 



Source link