State-Inference-Based Prompting for Natural Language Trading with Game NPCs

2025-07-09

AI Summary

Large language models (LLMs) frequently violate trading rules in game NPC systems—exhibiting item hallucination, erroneous price computation, and inconsistent state transitions—thereby undermining interaction reliability. To address this, we propose the State-Informed Behavioral Prompting (SIBP) framework, which explicitly models the trading process as six verifiable, sequential states. SIBP integrates context-aware item reference resolution with a placeholder-driven dynamic pricing mechanism, enabling lightweight, rule-compliant natural language transaction reasoning without fine-tuning or external modules. Evaluated on 100 test dialogues, SIBP achieves 97.2% state compliance, 95.4% reference accuracy, and 99.7% computational precision—substantially outperforming baseline methods. Our key contributions are threefold: (1) the first decomposition of trading logic into an auditable, state-transition flow; (2) a structured prompting design that jointly ensures correctness, interpretability, and low computational overhead; and (3) empirical validation of robust, zero-shot rule adherence in interactive game economies.

Technology Category

• Natural Language Processing: Prompt Engineering / Prompting
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Multiagent Systems: Mechanism Design

Application Category

• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

Large Language Models enable dynamic game interactions but struggle with rule-governed trading systems. Current implementations suffer from rule violations, such as item hallucinations and calculation errors, that erode player trust. Here, State-Inference-Based Prompting (SIBP) enables reliable trading through autonomous dialogue state inference and context-specific rule adherence. The approach decomposes trading into six states within a unified prompt framework, implementing context-aware item referencing and placeholder-based price calculations. Evaluation across 100 trading dialogues demonstrates >97% state compliance, >95% referencing accuracy, and 99.7% calculation precision. SIBP maintains computational efficiency while outperforming baseline approaches, establishing a practical foundation for trustworthy NPC interactions in commercial games.

Problem

• Enables reliable trading with NPCs using autonomous state inference
• Reduces rule violations like item hallucinations and calculation errors
• Improves trading dialogue accuracy and computational efficiency

Innovation

• State-Inference-Based Prompting for reliable trading
• Decomposes trading into six unified states
• Context-aware referencing and placeholder calculations