Illustration: Brian Hubble; portrait: Francis George

"From time to time I'll get a call where they say, 'You should be a proud American.'" —Jeffrey Jonas, database daddy-o

After building a database to keep track of a million dollars' worth of fish in the MGM Mirage Grand Hotel and Casino's gigantic aquarium, Jonas got tapped to help the casinos tackle a more daunting problem: how to keep out the bad guys. There was much at stake. A casino can lose its license if it's found doing business with anyone the Gaming Control Board has put on its exclusionary list. The problem is that casinos can't always figure out who is naughty or nice. Back then, there was simply no existing database program that could help them. As a result, casino security employees were leafing through mug-shot books by hand, trying to ferret out crooks on their own. It was time-consuming and inefficient. Jonas designed a solution—and proceeded to build a complete system in just 90 days.

The technology became known as NORA, short for Non-Obvious Relationship Awareness. NORA uses industry-standard relational databases that organize data into rows and columns for cross-referencing. As Jonas explains, there are essentially three piles of data for the casino to wade through: known cheats, ordinary players, and casino employees. Simply putting the three sets of data into three separate databases would make it difficult, if not impossible, to determine when and if any parties were colluding.

Instead, NORA ingests data from different data sets in multiple databases and combines them into one database using eXtensible Markup Language (XML). Whereas XML's predecessor, Hypertext Markup Language (HTML), identifies and codes a document's basic style elements such as an article's <title> or a <table>, XML lets programmers categorize and code not just those style elements but also items such as dates, prices, names, and locations.

Working in real time, NORA receives XML records from source systems and determines how these records are related to previously observed records. The benefits of this real-time processing allow the system to indicate when, for example, someone tagged as an excluded person makes a hotel reservation. More important, NORA can indicate whether people are who they say they are, as well as who is related to whom, even when they try to mask that information.

NORA, now marketed as IBM Relationship Resolution, begins processing by analyzing, categorizing, and notating—or standardizing—the data elements found in each record. For example, the names "Ricky" and "Dick" are noted to have the same root, "Richard." The program can compare addresses with postal base tables, in much the same way as direct marketing organizations ensure that addresses are valid. Records can also be enhanced by, say, adding latitude and longitude based on the street address. Once the record has been standardized and enhanced, the program evaluates the similarities and differences among entities—usually people or organizations—to determine if they are the same. For example, Bill Smith and William Smith at the same address and phone number might be identified as the same person, unless their dates of birth were different, which could indicate a father and son. After identities are resolved to determine who's who, the program shows how identities are interconnected.

In NORA's case, key data—Social Security numbers, names, birth dates, addresses—are recognized as features. By analyzing and matching these features, connections are made that might otherwise escape casino security staff. For example, when someone is employed as a dealer, NORA could take that individual's personal information—Social Security number, address, and so on—and compare it with other individuals' records. Also, because crooks often use different identities, NORA's meticulous cross-referencing means a casino won't miss Billy the Kid, Social Security number 555-55-4124, date of birth 11/6/50, when he identifies himself as The Kid, 555-55-2144, 6/11/50. "That's a different way of discovery," says Jonas. "It's valuable for [catching] people who are morphing identities." The end result might reveal that, say, the blackjack dealer's roommate is the leader of the biggest card-counting team in the country.

It was a no-brainer for the casinos; many of them have signed on, for a fee that Jonas declines to reveal. Meanwhile, Jonas has been fostering a related business with Griffin Investigations Inc., a detective agency that protects casinos from cheaters.