Enhancing Persuasive Dialogue Agents by Synthesizing Cross-Disciplinary Communication Strategies

TL;DR

A cross-disciplinary approach was applied to develop a framework for designing persuasive dialogue agents that draws on proven strategies from social psychology, behavioral economics, and communication theory and excelled at persuading individuals with initially low intent.

Abstract

Current approaches to developing persuasive dialogue agents often rely on a limited set of predefined persuasive strategies that fail to capture the complexity of real-world interactions. We applied a cross-disciplinary approach to develop a framework for designing persuasive dialogue agents that draws on proven strategies from social psychology, behavioral economics, and communication theory. We validated our proposed framework through experiments on two distinct datasets: the Persuasion for Good dataset, which represents a specific in-domain scenario, and the DailyPersuasion dataset, which encompasses a wide range of scenarios. The proposed framework achieved strong results for both datasets and demonstrated notable improvement in the persuasion success rate as well as promising generalizability. Notably, the proposed framework also excelled at persuading individuals with initially low intent, which addresses a critical challenge for persuasive dialogue agents.

Figures (4)

• Figure 1: Overview of the proposed framework. To simulate real-world persuasive dialogues, we deploy persuasive dialogue agents equipped with an extended set of empirically validated persuasive strategies (illustrated here for the P4G case).
• Figure 2: Proportion of cases in which each strategy increased the intention of persuadees to donate at each donation intention level.
• Figure 3: Proportion of Improved Donation Intentions after Strategy Use by Initial Donation Intention Level in English.
• Figure 4: Distribution of domains in the dataset constructed for the DailyPersuasion dataset.