Separation and classification is the process of forming a collection of loose pages into organized documents. In document separation you distinguish one group of pages as a distinct document and another group of pages as another document. Then, there's something that makes one kind of document different from another kind of document. An invoice is not an income statement. They are two different kinds of documents conveying two different sets of information. Classification is organizing documents into different groups, based on their content.
Before we get into how Grooper does that, ask yourself how would you do that in the "real world"? If you had a box of paper, how would you take those pages and turn them into documents? How would you determine one document is different from another? Ultimately, how you do this in the physical world and how Grooper digitizes that process will be remarkably similar.