Delegated updates in epistemic graphs for opponent modelling
|BIPOLAR; Computer Science; Computer Science, Artificial Intelligence; Epistemic argumentation; Epistemic updates; LOGIC; PROBABILISTIC ARGUMENTATION; RANKING; REVISION; SEMANTICS
|ELSEVIER SCIENCE INC
|INTERNATIONAL JOURNAL OF APPROXIMATE REASONING
In an epistemic graph, belief in arguments is represented by probability distributions. Furthermore, the influence that belief in arguments can have on the belief in other arguments is represented by constraints on the probability distributions. Different agents may choose different constraints to describe their reasoning, thus making epistemic graphs extremely flexible tools. A key application for epistemic graphs is modelling participants in persuasion dialogues, with the aim of modelling the change in beliefs as each move in the dialogue is made. This requires mechanisms for updating the model throughout the dialogue. In this paper, we introduce the class of delegated update methods, which harness existing, simpler update methods in order to produce more realistic outputs. In particular, we focus on hypothesized updates, which capture agent's reluctance or susceptibility to belief updates that can be caused by certain factors, such as time of the day, fatigue, dialogue length, and more. We provide a comprehensive range of options for modelling different kinds of agents and we explore a range of properties for categorising the options. (C) 2019 Elsevier Inc. All rights reserved.
Show full item record