Simply put, organizations can use Sensitive information types to classify data as sensitive. This includes addresses, credit card numbers, bank accounts, and more. Sensitive information types use patterns to identify certain information that may be sensitive and classify them using what is known as a confidence level. But what makes up a sensitive information type? A Sensitive information type is defined by a name, a description, and a pattern. The pattern determines what a sensitive information type is meant to detect.

Each pattern alongside the primary element has supporting elements, the confidence level, and the proximity. The supporting elements are effectively supporting evidence when the sensitive information type believes it found a match. The confidence level is the level of confidence that the matched information is a positive match, determined with a mixture of the primary element and supporting elements. The proximity is simply the amount of characters between primary and supporting elements.

Microsoft provides hundreds of different built-in sensitive information types to use. Each built-in sensitive information type is built using the primary and secondary elements but they utilize a variation of characteristics to describe them based on the information they are looking for. The defining components that are used to describe the built-in sensitive information types are format, pattern, checksum, keywords, and definition.

The format is the basic description of the sensitive information type. The pattern specifically defines that information. Depending on the types of information an organization may be handling, this could produce multiple positive matches which is why it's important to include the other 3 elements within sensitive information types whenever possible.

Sensitive information types also have the potential to add regular expression validators as a form of error detection, and a checksum is an example of one of these validators. Keywords are a way to further increase the likelihood of a positive match.

The definition is the confidence level that the sensitive information type has been detected based on the primary and supporting elements of the sensitive information type. This confidence level is displayed in a percentage which can be quickly used to determine validity of a positive match.

Now that we understand the main elements that create a sensitive information type, let's jump into Microsoft Purview and the Data classification solution within.