The main purpose of NAF is to provide a standardized format that is flexible enough to be used in a variety of NLP architectures. It should facilitate integration of tools developed at different sites and support a wide range of NLP applications, so that it can be used in current as well as future projects.

NAF is developed as part of NewsReader and BiographyNet. The format is based on formats used in other projects, such as KYOTO and OpeNER (KAF, Bosma et al., 2009) and TERENCE (TAF, Moens et al., 2011). It follows the basic strategy of the Linguistic Annotation Framework (LAF) as outlined by Ide et al. (2003). Furthermore, NAF uses URIs and has an alternative representation (RDF-NAF) that conforms to RDF. In this respect, it resembles the NLP Interchange Format (Hellman et al., 2013, NIF).