DKAL 2 — A Simplified and Improved Authorization Language
Yuri G. Gurevich, Itay Neeman · 2009
Knowledge and information are central notions in DKAL, a logic based authorization language for decentralized systems, the most expressive among such languages in the literature. Pieces of information are called infons. Here we present DKAL 2, a surprisingly simpler version of the language that expresses new important scenarios (in addition to the old ones) and that is built around a natural logic of infons. Trust became definable and its properties, postulated earlier as DKAL house rules, are now proved. In fact, none of the house rules postulated earlier is now needed. We identify also a most practical fragment of DKAL where the query derivation problem is solved in linear time. Note (added on May 11, 2009) In the meantime, we made the following notational changes. 1. Function put of type Info→Speech is renamed to implied. The new notation makes it obvious that (p implied x) is weaker than (p said x). 2. The conjunction of infons x, y is denoted as x ∧ y rather than x + y. While the latter notation made sense in the original algebraic approach, it is not natural in the logic approach. Contents