Via document classification, a document is automatically assigned to a category, out of a large set of predefined categories. For example, the document can be about "Sport", or "Science and Technology" or "Politics", etc. A document can actually belong to multiple categories, with higher or lower affinity.
Technically, the task of document classification is carried out by a machine learning model, trained on a large corpus of example documents, from a predefined categorization scheme. The definitions of categories are inherently domain-specific, as it is hard to define a scheme that encompasses "all themes" that text documents can be about. We opted for a categorization that best suits news data (also suitable to blogs, twitter, etc.). The technical documentation presents the approach in detail.