BaZi-Based Character Simulation Benchmark: Evaluating AI on Temporal and Persona Reasoning

2025-10-27

AI Summary

Current virtual character generation faces bottlenecks including heavy reliance on annotated data, weak personality coherence, and insufficient cultural adaptation. To address these, we propose BaZi-LLM—a novel framework that pioneers the integration of Chinese Bazi (Eight Characters) astrology as a structured personality constraint mechanism, synergizing symbolic reasoning with large language models (e.g., DeepSeek-v3, GPT-5-mini). We further introduce the first Bazi-oriented QA dataset grounded in life domains—wealth, health, family, career, and relationships. Our approach enables fine-grained, temporally dynamic virtual personality modeling, significantly enhancing character authenticity and cultural consistency. Evaluation across multiple dimensions shows accuracy improvements of 30.3%–62.6% over baselines. Moreover, performance degrades by 20%–45% under erroneous Bazi inputs, confirming deep cultural logic embedding and robust reasoning consistency. This work establishes a new paradigm for culture-enhanced AI character modeling.

Technology Category

• Humans and AI: Game Design — Virtual Humans, NPCs and Autonomous Characters
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Human-like virtual characters are crucial for games, storytelling, and virtual reality, yet current methods rely heavily on annotated data or handcrafted persona prompts, making it difficult to scale up and generate realistic, contextually coherent personas. We create the first QA dataset for BaZi-based persona reasoning, where real human experiences categorized into wealth, health, kinship, career, and relationships are represented as life-event questions and answers. Furthermore, we propose the first BaZi-LLM system that integrates symbolic reasoning with large language models to generate temporally dynamic and fine-grained virtual personas. Compared with mainstream LLMs such as DeepSeek-v3 and GPT-5-mini, our method achieves a 30.3%-62.6% accuracy improvement. In addition, when incorrect BaZi information is used, our model's accuracy drops by 20%-45%, showing the potential of culturally grounded symbolic-LLM integration for realistic character simulation.

Problem

• Creating benchmark for temporal persona reasoning evaluation
• Developing symbolic-LLM integration for dynamic character simulation
• Addressing cultural grounding in virtual persona generation

Innovation

• Integrates symbolic reasoning with large language models
• Generates temporally dynamic and fine-grained virtual personas
• Uses culturally grounded BaZi-based persona reasoning dataset