A propagation-based ranking semantics in Explainable Bipolar Weighted Argumentation

The Semantic Web of Things enables exchanging knowledge-based fragments in pervasive computing through machine-to-machine interactions, sustaining collaborative decision-making in networks of smart objects. Abstract argumentation is widely acknowledged, instead, as a general framework for decision-making in multi-agent systems, able to solve negotiation conflicts and evaluate acceptable options for a given scenario. Nevertheless, abstract argumentation frameworks disregard the problem of characterizing the argument structure and their mutual relations. This paper proposes a novel approach integrating Dung-style abstract argumentation with semantic matchmaking in pervasive computing. Object interactions are seen as an argumentative dialogue, where annotations shared by each agent play the role of arguments. A matchmaking scheme, exploiting non-standard non-monotonic inferences, allows the appraisal of argument relations in a bipolar weighted argumentation approach. A propagation-based gradual semantics provides acceptability ranking of arguments with a formal explanation. A fading mechanism is also exploited to take into account limited computational resources. As the proposed approach enables applications in cyber-physical multi-agent systems engineering, early validation and performance experiments are provided by means of a case study on a real-time strategy game.