Problem
Research questions and friction points this paper is trying to address.
credit card simulation
online discussions
realistic dialogue generation
user interaction modeling
conversational realism
CARD: Controlled Agentic Reddit Discussions for Credit Card Simulation
🤖 AI Summary
Existing methods struggle to generate credit card discussion posts that authentically replicate both user expression and interaction patterns, often lacking fine-grained control over multidimensional attributes such as structure, stance, and tone. This work proposes CARD, a novel framework that uniquely integrates structured planning with distribution calibration. CARD employs a planner to specify non-token-level guidance signals—including reply structure, function, stance, tone, and diversity—which a writer module then uses to generate content. A calibration loop iteratively refines the output distribution to better align with real-world data. Evaluated across multiple large language models, CARD significantly outperforms baseline approaches, producing synthetic discussions that more closely match authentic Reddit credit card threads in terms of thread-level distribution, effect size, and distributional distance, thereby enabling controllable generation of financial dialogues with rich, multidimensional fidelity.
📝 Abstract
Online credit card discussions provide a natural setting for studying how consumers communicate about financial products. Simulating these discussions requires more than just generating individual comments, the generated threads should also match how real users express themselves and interact with others. We introduce CARD, a framework for generating realistic credit card discussion threads. Given a credit card post and its matched real thread, CARD uses non-verbatim guidance on reply structure, comment function, stance, tone, and conversational variation. A planner organizes these controls, a writer generates the discussion, and a calibration loop updates comments' populations that contribute to differences between the generated and real thread distributions. We evaluate CARD on real Reddit credit card discussions using lexical, semantic, behavioral, and structural metrics. CARD matches the distributions of real credit card discussions better than simulation baselines across multiple LLMs and also demonstrates smaller effect sizes and distribution distances across metrics. These results show that structured planning and targeted revision can generate the realism of simulated credit card discussions.
Innovation
Methods, ideas, or system contributions that make the work stand out.
controlled generation
conversational simulation
distribution calibration
structured planning
Reddit discussion modeling
🔎 Similar Papers
No similar papers found.
