* IBM DB2 Anonymous Resolution
On May 17, an Alitalia flight from Italy bound for Boston was diverted to Bangor, Maine because a passenger’s name was a match for a name on the U.S. government’s no-fly list. The passenger was removed from the flight and questioned by the FBI, which said he is not a terrorist. The passenger was not arrested.
This was a near repeat of an incident on May 12, in which a Boston-bound Air France flight was diverted to Bangor because a passenger had the same name and birth date as a person on the no-fly list. The passenger, along with a woman and two children, was removed from the flight, questioned and then released.
Airlines are required by U.S. law to provide the Homeland Security Department with a passenger list for all flights headed to the U.S. within fifteen minutes of takeoff. The process of comparing the passenger list to the no-fly list is imperfect and yields a lot of costly false positives, as in the two cases in May.
The process could be improved if the airlines and the Homeland Security Department agreed to use a new product from IBM called DB2 Anonymous Resolution. This software allows two organizations to anonymously share and compare sensitive data without revealing personal details from either list. It’s a solution that could work well not just in the transportation industry, but also in the financial services, hospitality, and health care sectors, and in law enforcement and even mergers and acquisitions.
DB2 Anonymous Resolution is part of IBM’s family of products called IBM DB2 Entity Analytic Solutions. They were developed by Systems Research & Development, which was acquired by IBM earlier this year. The product family includes DB2 Identity Resolution (formerly called ERIK), DB2 Relationship Resolution (formerly called NORA), and the newly-released DB2 Anonymous Resolution. Since the products’ capabilities build on each other, I’ll take a moment to describe them all.
DB2 Identity Resolution analyzes streams of data from multiple sources and creates an accurate record of “who is who.” For example, let’s say we are building that no-fly list for the first time. The government might take data from sources such as criminal records, passport records, immigration records, and so on. The software would analyze all the records and create a unique entity identifier for a specific person, despite that person being represented differently in each database. He’s John Doe in one database, Jonathon Doe in another, and J. Doe-Smith in a third. In the end, he is entity #187654.
Now, that no-fly list might be made more meaningful if we not only look at who someone is, but also who he knows. While John Doe may not be worthy of the list by his own right, the fact that he once shared an apartment with Osama Bin Laden would certainly make him worth watching closely. DB2 Relationship Awareness software can tell us “who knows who.” It detects non-obvious as well as obvious relationships, and enables instantaneous alerts when relationships are discovered.
That brings us to the next level: “Who is who and who knows who…anonymously.” Air France knows quite a few details about who is on one of its planes, including frequent flyer histories, credit card numbers, and so on. The airline might be reluctant to yield that data to the U.S. government. For one thing, it could be a violation of European Union privacy laws. Likewise, the U.S. government doesn’t want to send its no-fly list in the clear to Air France to compare to its passenger list.
Both organizations, however, could take their separate lists and anonymize them with DB2 Anonymous Resolution’s one-way hashing algorithm, which creates a unique digital signature for each record. Now you can apply a matching function between the two lists without ever revealing private data to the other party. If there is a hit, you can issue an alert back to the airline and Homeland Security and act before the plane finishes taxiing down the runway.
Obviously the more data records and sources you feed into such systems, the better your analysis can be. The DB2 Entity Analytic Solutions are designed for real-time high volume processing. According to IBM, one customer has more than 5,000 data sources accessing over 500 million records, with 2,200 entities resolved per second on the entire U.S. population. Data sources can be internal or external, fed as XML files.
Potential business uses include:
* Getting a true look at the overlap of customers in a merger situation.
* Reducing the risk of sharing customer records with foreign subsidiaries.
* Performing statistical analysis on medical records to analyze incidence rates of diseases and other conditions.
DB2 Anonymous Resolution can tame organizations’ reluctance to share data by reducing the risk of unintended disclosure. And maybe it can help prevent some of those planes from getting diverted to Bangor.
Linda Musthaler is vice president of Currid & Company. You can write to her at mailto:Linda.Musthaler@currid.com
