Semantic Modelling of Plans and Execution Traces for Enhancing Transparency of IoT Systems

Milan Markovic, Daniel Garijo, Peter Edwards, Wamberto Vasconcelos

Abstract: Transparency of IoT systems is an essential requirement for enhancing user's trust towards such systems. Provenance mechanisms documenting the execution of IoT systems are often cited as an enabler of such transparency. However, provenance records often lack detailed descriptions of a system's expected behaviour. Plan specifications describe the steps needed to achieve a certain goal by a human or an automated system. Once plans reach a certain level of complexity, they are typically decomposed in different levels of abstraction. However, this decomposition makes it difficult to relate high level abstract plans to their granular execution traces. This paper introduces EP-Plan, a vocabulary for linking the different levels of granularity of a plan with their respective provenance traces. EP-Plan also provides the means to describe plan metadata such as constraints, policies, rationales, and expected participating agents associated with a plan.