跳转到主要内容

关于VULCA About VULCA

AI艺术评论实验系统 AI-Powered Art Criticism Experimental System

VULCA 系统简介 VULCA System Overview

VULCA（跨时代艺术评论视角）是一个基于人工智能的艺术评论实验系统，旨在探索不同历史时期的艺术评论家如何评价当代艺术作品。 VULCA (Cross-Temporal Art Criticism Perspectives) is an AI-powered art criticism experimental system designed to explore how art critics from different historical periods evaluate contemporary artworks.

系统采用AI角色扮演技术，模拟6位来自不同时代和文化背景的艺术评论家，对4件当代艺术作品进行多维度评论。每位AI评论家基于历史文献和理论文本进行角色建模，在RPAIT五维框架下生成评论，形成6×4×5=120个评论数据点的实验数据集。 The system employs AI role-playing technology to simulate 6 art critics from different historical periods and cultural backgrounds, providing multi-dimensional critiques of 4 contemporary artworks. Each AI critic is modeled based on historical documents and theoretical texts, generating critiques under the RPAIT five-dimensional framework, forming an experimental dataset of 6×4×5=120 critique data points.

通过这一系统，我们探索以下研究问题：历史上的艺术评论家会如何看待当代艺术？AI能否准确模拟历史人物的评论风格和思想体系？结构化的评论框架能否帮助我们更系统地理解艺术作品？ Through this system, we explore the following research questions: How would historical art critics view contemporary art? Can AI accurately simulate the critical style and ideological systems of historical figures? Can a structured critique framework help us understand artworks more systematically?

RPAIT 评论框架 RPAIT Critique Framework

RPAIT是VULCA评论的核心框架，由五个关键维度组成： RPAIT is the core framework of VULCA critiques, consisting of five key dimensions:

R - 代表性 R - Representation

作品如何呈现现实或观念？它选择了什么，又回避了什么？代表性涉及艺术家对现实的选择性解读和表达方式。 How does the work present reality or concepts? What does it select, and what does it avoid? Representation involves the artist's selective interpretation and expression of reality.

P - 哲学性 P - Philosophicality

作品所体现的哲学思想和世界观是什么？它涉及什么样的人类根本问题？哲学性关乎作品的理念深度。 What philosophical thoughts and worldviews does the work embody? What fundamental human questions does it address? Philosophicality concerns the conceptual depth of the work.

A - 美学性 A - Aesthetics

作品的视觉形式、颜色、构图、材料等如何创造美的体验？美学性是艺术最直接的感官维度。 How do the work's visual form, color, composition, and materials create aesthetic experience? Aesthetics is art's most direct sensory dimension.

I - 诠释性 I - Interpretability

作品能够被如何诠释？它开放了多少种可能的意义？诠释性反映了作品的意义开放性和观众的参与空间。 How can the work be interpreted? How many possible meanings does it open? Interpretability reflects the work's semantic openness and audience participation space.

T - 技巧性 T - Technicality

艺术家运用了什么样的技法和表现手段？技巧的娴熟度如何？技巧性体现了艺术执行的专业性。 What techniques and expressive means does the artist employ? How skillful is the execution? Technicality embodies the professionalism of artistic execution.

研究目标 Research Objectives

VULCA系统的核心研究目标包括以下三个方面： The core research objectives of the VULCA system include the following three aspects:

通过这些研究，我们希望为AI辅助的艺术评论、历史人物角色建模以及跨时代文化对话提供新的视角和方法。 Through these studies, we aim to provide new perspectives and methods for AI-assisted art criticism, historical figure persona modeling, and cross-temporal cultural dialogue.

系统架构 System Architecture

VULCA系统由三个核心组件构成： The VULCA system consists of three core components:

AI评论家角色系统 AI Critic Persona System

基于历史文献和理论文本，为每位评论家建立角色模型，包括其美学观点、哲学立场、文化背景和评论风格。系统采用 Based on historical documents and theoretical texts, establish persona models for each critic, including their aesthetic views, philosophical stances, cultural backgrounds, and critical styles. The system employs a mixed design of 4位真实历史人物 4 real historical figures 和 and 2位虚构AI角色 2 fictional AI personas 的混合设计： :

真实历史评论家（4位） Real Historical Critics (4) ：苏轼（北宋文人，1037-1101）、郭熙（北宋山水画家，1020-1100）、约翰·罗斯金（维多利亚时期评论家，1819-1900）、凯特·克劳福德（当代AI伦理学家）。这些评论家的理论立场和著作均有历史文献支撑。 : Su Shi (Northern Song literatus, 1037-1101), Guo Xi (Northern Song landscape painter, 1020-1100), John Ruskin (Victorian critic, 1819-1900), Kate Crawford (contemporary AI ethicist). These critics' theoretical positions and works are supported by historical documents.

虚构AI角色（2位） Fictional AI Personas (2) ：佐拉妈妈（代表西非griot口述传统和集体诠释视角）、埃琳娜·佩特洛娃教授（代表俄罗斯形式主义结构分析传统）。这两位虚构角色由AI创建，用于代表特定的批评范式和文化视角，是本研究探索AI角色扮演能力的实验设计。 : Mama Zola (representing West African griot oral tradition and collective interpretive perspective), Professor Elena Petrova (representing Russian formalist structural analysis tradition). These two fictional personas are created by AI to represent specific critical paradigms and cultural perspectives, serving as experimental design for exploring AI role-playing capabilities in this research.

这种真实与虚构相结合的设计，既保证了历史评论传统的准确性，又探索了AI在构建新型批评视角时的创造潜力。 This combined design of real and fictional ensures both the accuracy of historical critical traditions and explores AI's creative potential in constructing novel critical perspectives.

RPAIT评论框架 RPAIT Critique Framework

五维度结构化评论系统，确保评论的全面性和可比性。每个维度都有明确的评估标准：代表性关注作品对现实的呈现方式，哲学性关注思想深度，美学性关注视觉形式，诠释性关注意义的开放性，技巧性关注艺术执行的专业性。每个维度采用1-10分量化评分。 A five-dimensional structured critique system ensuring comprehensiveness and comparability of critiques. Each dimension has clear evaluation criteria: Representation focuses on how the work presents reality, Philosophicality on conceptual depth, Aesthetics on visual form, Interpretability on semantic openness, and Technicality on professional execution. Each dimension uses a 1-10 quantified scoring system.

评论生成管道 Critique Generation Pipeline

结合AI生成和人工审核的混合流程。首先通过大语言模型基于角色设定和RPAIT框架生成初稿评论，然后由研究团队进行历史准确性验证、理论一致性检查和文本质量控制，最终形成符合学术标准的评论数据集。 A hybrid workflow combining AI generation and human review. First, large language models generate initial critiques based on persona settings and the RPAIT framework, then the research team performs historical accuracy verification, theoretical consistency checks, and text quality control, ultimately forming an academically rigorous critique dataset.

研究意义 Research Significance

VULCA系统在学术研究和实际应用方面都具有重要意义： The VULCA system holds significant importance in both academic research and practical applications:

学术贡献 Academic Contributions ：本研究为艺术评论的计算方法、历史人物的AI建模以及跨时代文化对话提供了新的研究范式。RPAIT框架为艺术评论的结构化分析提供了可复用的工具，而AI角色扮演技术为人文研究中的"反事实历史"探索开辟了新路径。 : This research provides new paradigms for computational methods in art criticism, AI modeling of historical figures, and cross-temporal cultural dialogue. The RPAIT framework offers reusable tools for structured analysis of art criticism, while AI role-playing technology opens new paths for exploring "counterfactual history" in humanities research.

应用场景 Application Scenarios ：系统的方法和框架可应用于艺术教育（帮助学生从多维度理解艺术）、策展实践（为展览提供多元评论视角）、艺术批评研究（系统分析不同批评传统的特点）以及AI人文应用（探索AI在历史人物建模和文化传承中的可能性）。 : The system's methods and frameworks can be applied to art education (helping students understand art from multiple dimensions), curatorial practice (providing diverse critical perspectives for exhibitions), art criticism research (systematically analyzing characteristics of different critical traditions), and AI humanities applications (exploring AI's potential in historical figure modeling and cultural transmission).

更重要的是，VULCA系统提出了一个关键问题：当我们让AI"代言"历史人物时，我们究竟在创造什么？是对历史的再现，还是一种新的诠释？这一问题对于理解AI在人文领域的作用至关重要。 More importantly, the VULCA system raises a crucial question: When we let AI "represent" historical figures, what exactly are we creating? Is it historical reproduction or a new interpretation? This question is crucial for understanding AI's role in the humanities.

相关论文 Related Publication

A Structured Framework for Evaluating and Enhancing Interpretive Capabilities of Multimodal LLMs in Culturally Situated Tasks

Haorui Yu, Ramon Ruiz-Dolz, Qiufeng Yi

Findings of the Association for Computational Linguistics: EMNLP 2025, pages 1945-1971, Suzhou, China

BibTeX 引用 BibTeX Citation 
@inproceedings{yu-etal-2025-structured,
    title = "A Structured Framework for Evaluating and Enhancing Interpretive Capabilities of Multimodal {LLM}s in Culturally Situated Tasks",
    author = "Yu, Haorui  and
      Ruiz-Dolz, Ramon  and
      Yi, Qiufeng",
    editor = "Christodoulopoulos, Christos  and
      Chakraborty, Tanmoy  and
      Rose, Carolyn  and
      Peng, Violet",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
    month = nov,
    year = "2025",
    address = "Suzhou, China",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2025.findings-emnlp.103/",
    pages = "1945--1971",
    ISBN = "979-8-89176-335-7"
}

联系我们 Contact Us

对VULCA项目有任何问题或建议？欢迎联系： Have any questions or suggestions about the VULCA project? Feel free to contact:

yuhaorui48@gmail.com