One way to fight white-collar crime and thwart terror is to identify who is transferring money through banks around the world. Stiff new laws in the European Union are forcing banks to check whether funds could be linked to lists of suspected terrorists or other criminals. But the banks are often stumped by the varied spellings of names in different countries.

Now, a Virginia company that spent years developing a massive name-matching database for U.S. intelligence agencies is selling its system to help banks fight crime. The database from Language Analysis Systems of Herndon, VA, is a fundamental tool that helps reconcile myriad variations on names. And it can help determine whether someone on a watch list is the person who holds a bank account. The database contains one billion names from 200 countries and is regularly updated. But numbers alone don't do the job: "People needed better help understanding names," says Jack Hermansen, CEO of Language Analysis. Software places emphasis on different portions of names according to cultural rules. For instance, it will filter out a modifier on Arabic names that only indicates whether or not a person made a pilgrimage to Mecca, and it will disregard long common endings in Greek names. The company continually refines such strategies. By comparing only key name elements, the software finds better matches.