Preserving digital objects requires preservation of not only their bit-level representation but also their intelligibility. To this end a digital object should be associated with metadata appropriate for interpreting that object; such metadata are often referred as representation information. Even such metadata may not be intelligible, however, so we may have to associate them with extra metadata, and so on. This paper approaches this problem by introducing an abstract model comprising modules and dependencies. Community knowledge is formalized over the same model by introducing the notion of profile. This notion is then exploited for deciding representation information adequacy (during input) and intelligibility (during output). Subsequently some general dependency management services for identifying and filling intelligibility gaps during input and output are described and analysed. Finally a prototype system based on these ideas is described.