Classifications and customizations. Data classification in the context of Microsoft Purview is a way of categorizing data assets by assigning unique logical tags or classes to the data assets. Classification is based on the business context of the data. For example, we might classify assets as passport number, driver's license number, credit card number, or an email address. When we classify data assets, we make them easier to understand, search, and govern. Classifying data assets also helps us understand the risks associated with them. This, in turn, can help us implement measures to protect sensitive or important data from ungoverned proliferation and unauthorized access across the data state. 

Microsoft Purview provides an automated classification capability while we scan our data sources. We get more than 200 built-in system classifications and the ability to create custom classifications for our data. We can classify assets automatically when they're configured as part of a scan or we can edit them manually in the Microsoft Purview governance portal after they've been scanned and ingested. Classification can be particularly important for data governance. 

Among other reasons, classifying data assets is important because it helps: narrow down the search for data assets that we're interested in, organize and understand the variety of data classes that are important in our organization and where they're stored, understand the risks associated with our most important data assets and then take appropriate measures to mitigate them. There are two main types of classifications available in Purview. System classifications are the 200+ out of the box classifications already built into Purview — and include items like person names, email addresses, phone numbers, and passport numbers. Custom classifications are classifications that we create based on a pattern or a specific column name that's unavailable as a system classification. Custom classification rules can be based on a regular expression pattern or dictionary. We can also apply both types of classifications to file assets, table assets, and column assets.