Multi-Party Persuasion: a Paraconsistent Approach

Abstract

Some conflicts appearing in multi-agent settings may be resolved via communication. In this paper, besides conflicts of opinions, paradigmatically resolved by a persuasion dialogue, we study semantically deeper conflicts reaching to motivations of opinions. This investigation led us to discerning deep persuasion dialogues aiming at the resolution of conflicting motivations of opinions.

In our overall research program we focus on realistic modeling of agency. This includes a proper representation of agents’ ignorance and inconsistencies, appearing in their informational stance. Therefore, our formal framework TalkLOG, designed to provide and embed different forms of dialogues, employs a 4-valued logic with two additional logical values, unknown and inconsistent. Within TalkLOG soundness and completeness of persuasion was proved by comparing the outcomes of the persuasion dialogues performed by n-agents with the outcomes obtained by merging knowledge of these n agents. In this context the key point was a proper construction of this merge operator.

Another critical issue is complexity of agents’ communication, which is typically interleaved with reasoning in the context of multi-agent or autonomous systems. In TalkLOG tractability of both aspects is obtained thanks to the implementation tool: rule-based 4-valued query language 4QL.