Search papers, labs, and topics across Lattice.
The CARD framework was developed to generate realistic discussions about credit cards by simulating the nuanced interactions found in actual online forums. By employing non-verbatim guidance on various aspects of conversation, including stance and tone, CARD effectively captures the complexity of consumer communication. Evaluation against real Reddit discussions reveals that CARD outperforms existing simulation baselines in matching the distribution of real interactions across multiple metrics, highlighting the effectiveness of structured planning and iterative refinement in dialogue generation.
CARD achieves a level of realism in credit card discussions that outstrips traditional simulation methods, revealing the power of structured conversational planning.
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.
Stanford HAI