CVC: A Large-Scale Chinese Value Rule Corpus for Value Alignment of Large Language Models

2025-06-02

AI Summary

Current LLM value evaluation suffers from Western-centrism, absence of a Chinese-aligned value framework, and poor scalability in cross-cultural scenario generation. To address these issues, we propose the first hierarchical value system grounded in core Chinese values—comprising three dimensions, twelve core values, and fifty derived values—alongside CVC, a large-scale Chinese Value Corpus containing over 250K human-enhanced value rules and 400K moral dilemma scenarios. We introduce a rule-driven, cross-culturally sensitive scenario generation paradigm integrating hierarchical value modeling, rule-augmented annotation, adversarial generation of sensitive topics, and multi-model preference evaluation. Experiments demonstrate that CVC-guided scenarios significantly outperform baselines in value boundary fidelity and scenario diversity; seven mainstream LLMs select CVC-aligned options in over 70.5% of cases across six sensitive issue categories; and five Chinese annotators achieve an 87.5% agreement rate on CVC’s value expression validity.

Technology Category

• Humans and AI: Learning Human Values and Preferences
• Philosophy and Ethics of AI: Morality & Value-based AI
• Natural Language Processing: Safety and Robustness

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Ensuring that Large Language Models (LLMs) align with mainstream human values and ethical norms is crucial for the safe and sustainable development of AI. Current value evaluation and alignment are constrained by Western cultural bias and incomplete domestic frameworks reliant on non-native rules; furthermore, the lack of scalable, rule-driven scenario generation methods makes evaluations costly and inadequate across diverse cultural contexts. To address these challenges, we propose a hierarchical value framework grounded in core Chinese values, encompassing three main dimensions, 12 core values, and 50 derived values. Based on this framework, we construct a large-scale Chinese Values Corpus (CVC) containing over 250,000 value rules enhanced and expanded through human annotation. Experimental results show that CVC-guided scenarios outperform direct generation ones in value boundaries and content diversity. In the evaluation across six sensitive themes (e.g., surrogacy, suicide), seven mainstream LLMs preferred CVC-generated options in over 70.5% of cases, while five Chinese human annotators showed an 87.5% alignment with CVC, confirming its universality, cultural relevance, and strong alignment with Chinese values. Additionally, we construct 400,000 rule-based moral dilemma scenarios that objectively capture nuanced distinctions in conflicting value prioritization across 17 LLMs. Our work establishes a culturally-adaptive benchmarking framework for comprehensive value evaluation and alignment, representing Chinese characteristics. All data are available at https://huggingface.co/datasets/Beijing-AISI/CVC, and the code is available at https://github.com/Beijing-AISI/CVC.

Problem

• Addressing Western bias in LLM value alignment
• Lack of scalable Chinese value evaluation frameworks
• High-cost, inadequate cross-cultural scenario generation

Innovation

• Hierarchical Chinese value framework with 50 values
• Large-scale corpus with 250k human-annotated rules
• Rule-based moral dilemma scenarios for 17 LLMs