Information Technology Reference
In-Depth Information
Data mining
is the process of searching through one or more databases looking for
patterns or relationships among the data. In this section we continue our coverage of the
information-processing category of Solove's taxonomy by surveying a few well-known
data-mining projects run by government agencies.
To identify taxpayers who have paid less in taxes than they owe, the Internal Revenue
Service (IRS) uses computer-matching and data-mining strategies. First, it matches in-
formation on the tax form with information provided by employers and financial insti-
tutions. This is a straightforward way to detect unreported income.
Second, the IRS audits a couple of million tax returns every year. Its goal is to
select the most promising returns—those containing errors resulting in underpayment
of taxes. The IRS uses a computerized system called the discriminant function (DIF) to
score every tax return. The DIF score is an indicator of how many irregularities there are
on a tax form, compared to carefully constructed profiles of correct tax returns. About
60 percent of tax returns audited by the IRS are selected due to their high DIF scores.
Another application of data mining by the government is protecting society from immi-
nent dangers.
A syndromic surveillance system is a computerized system that analyzes 911 calls,
visits to the emergency room, school absenteeism, purchases of prescription drugs,
and Internet searches to find patterns that might indicate the onset of an epidemic, an
environmental problem leading to illnesses, or bioterrorism.
In the fall of 2002, a syndromic surveillance system in New York City detected a
surge in people seeking treatment for vomiting and diarrhea. These symptoms were the
first signs of an outbreak of a Norwalk-type virus. The alert generated by the system
allowed city officials to warn doctors about the outbreak and advise them to be particu-
larly careful about handling the highly contagious body fluids of their affected patients
[68].
Shortly after September 11, 2001, several major telecommunications providers began
turning over the phone call records of tens of millions of Americans to the National
Security Agency, without a court order. The NSA was not monitoring or recording the
actual conversations; instead, it was analyzing calling patterns in order to detect potential
terrorist networks. [69].
After
USA Today
revealed the existence of the database in May 2006, more than a
dozen class-action lawsuits were filed against the telecommunications companies. In Au-
gust 2006, a federal judge in Detroit ruled the program to be illegal and unconstitutional,
