How to Design FDO Profiles and Kernel Attributes?

A major goal for FDOs is to enable machine readability, interpretability and actionability for data and metadata of digital objects. This paper examines, how this can be achieved at the level of processing, grammars, derived languages and push-down automata, and describes formal requirements for the definition of FDO records. After a description of FDOs and what requirements are necessary that they can be processed by machines the importance of types for the processing of bitstreams as well as key-value-pairs in attributes is highlighted and the different options to represent values in attributes of FDOs are investigated. Machine actionability describes the knowledge of machines about how to process the object. This requires to process the grammars behind the languages that can be used for attribute definitions to determine the type of the value and the grammars that generate these languages. A way out of a possibly unlimited need for different machines able to read different languages is presented on the level of grammars. Even if these findings are on an abstract level, the outcome has direct consequences for the way how FDO records, data types or profiles have to be defined, which standardization agreements are needed and thus how attributes, types or profiles have to be implemented. It can be seen as a theoretical guideline for standarization and for implementation.