用AI分析唐宋诗词中的花意象情感，发现文学与艺术的深层共鸣
Flower Across Time and Media: Sentiment Analysis of Tang Song Poetry and Visual Correspondence
- 用微调BERT模型分析唐宋诗词中牡丹与梅花的情感
- 唐诗重豪放，宋词更含蓄，情感倾向有明显时代差异
- 发现诗词情感与同期纺织品、瓷器纹样高度一致
唐（618–907）与宋（960–1279）时期是中国文化表达的高峰，花卉意象成为抒发情感与艺术设计的重要载体。以往研究多将文学与视觉艺术分开探讨，二者之间的系统性关联仍不明确。本研究采用基于BERT的文本情感分析方法，量化唐宋诗词中牡丹与梅花意象的情感变化，并与同期纺织品、陶瓷等物质文化中的视觉表现进行交叉验证。结果表明，唐诗中花意象多体现豪放壮美，宋词则转向内省婉约，情感基调发生显著转变；这一变化在服饰纹样、器物装饰中亦可得到印证，揭示出文学表达与艺术创作之间深层次的同步演进关系。
原文摘要 · Abstract (English)
The Tang (618 to 907) and Song (960 to 1279) dynasties witnessed an extraordinary flourishing of Chinese cultural expression, where floral motifs served as a dynamic medium for both poetic sentiment and artistic design. While previous scholarship has examined these domains independently, the systematic correlation between evolving literary emotions and visual culture remains underexplored. This study addresses that gap by employing BERT-based sentiment analysis to quantify emotional patterns in floral imagery across Tang Song poetry, then validating these patterns against contemporaneous developments in decorative arts.Our approach builds upon recent advances in computational humanities while remaining grounded in traditional sinological methods. By applying a fine tuned BERT model to analyze peony and plum blossom imagery in classical poetry, we detect measurable shifts in emotional connotations between the Tang and Song periods. These textual patterns are then cross berenced with visual evidence from textiles, ceramics, and other material culture, revealing previously unrecognized synergies between literary expression and artistic representation.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。