Automating the Data Match Game

By David Rubinstein

April 1, 2008 — Poor Alan Rodgers. Sometimes he’s Allen Rodgers. Other times he’s Allan Rogers. Or Rodjers.

Although simply an annoyance most times, the misspellings could lead to something much worse. Imagine that Alan has to go into the hospital for an emergency procedure, but the doctor can’t find his record because no matter how he types in the record request, he can’t match the name to the desired information. Now, the doctor might not know that Alan is allergic to a certain anesthesia, or that he has high blood pressure that could be exacerbated by certain medications.

Data shared in databases is imperfect and rife with discrepancies. And, according to Stef Damianakis, president and CEO of data management tool provider Netrics, is not an effective tool for matching up data that is similar but not identical. He cited the sources of imprecise data: humans, who make mistakes entering the data, and computers that speak to each other but can’t recognize similarities in entries and decide if the entries are in fact representing the same thing.

Netrics is trying to automate the problem of matching data requests to records when the names are close, by quantifying what “close” means in mathematical terms. The company has released version 4 of its Data Matching Platform that includes a matching engine that uses algorithmic expressions to find matches even if the similarity is incomplete. Damianakis gave the example of two records that have matching gender entries but similar first names and different last names, noting that those entries could represent a woman who took a married name. There is also a decision engine that detects hidden patterns in data and learns to draw conclusions from those patterns, according to Damianakis.

The Netrics platform utilizes sophisticated pattern matchers to look at records from the field level and then from a higher view, looking for patterns that might connect two records. “The problem is that expertise in someone’s head is hard to get into a computer program,” he said. “Ask an expert to describe the steps he takes to balance his checkbook, to put it down in English, and an engineer can take that and code it up. But other problems can’t be done this way. An expert can’t write down, for example, the steps he takes to recognize a voice. But everyone does it.”

Version 4 of the platform, released in late February, provides Web services capability that Damianakis said makes it more enterprise-ready. SOAP support, native C/, Java, .NET and Python APIs and real-time data loading and synch round out the offering.

Related Search Term(s): Data matching, databases

Share this link: https://sptechweb.com/link/31919

The Data Center: When Setting Up Data Center, Don't Forget the Network! You can build the biggest, most advanced data centers on the planet, but as Google and Microsoft know, without a network, they are not too useful. Companies have been investing in expanding global network capacity to this end.

The Data Center: Using the Three P's of Disaster Recovery to Protect Your Data Policy, process and procedures are the three legs of disaster recovery. Using backup data centers in remote locations, or by using containers, can prevent crippling losses of data and can speed up recovery after a catastrophe.

CA Adds Data Loss Prevention With Orchestria Purchase CA executives said Orchestria’s data loss prevention technology complements the company’s server protection capabilities and prevents data breaches. This is the third security-related acquisition in the past six months for CA.

Add comment

Name*
Email*
Country

Comment
Preview

Automating the Data Match Game

By David Rubinstein

Related Articles

Add comment