关于这个项目
About This Project
梵高一生写了 897 封信，画了 1030 幅画。这个项目试图用数据重建他内心状态的轨迹——不是为了解释他，而是为了更靠近他。
Vincent van Gogh wrote 897 letters and created 1,030 paintings over the course of his life. This project attempts to reconstruct the arc of his inner states through data — not to explain him, but to draw closer to him.
-
信件Letters
Van Gogh Letters Project（vangoghletters.org）英译版全集，收录梵高写给 Theo、Anton van Rappard、Paul Gauguin 等人的全部 897 封信，涵盖 1872–1890 年。
Complete English translation from the Van Gogh Letters Project (vangoghletters.org). Includes all 897 letters written by Van Gogh to Theo, Anton van Rappard, Paul Gauguin, and others, spanning 1872–1890.
-
画作Paintings
Wikimedia Commons，附带认可度评分——基于博物馆馆藏等级（梵高博物馆、奥赛博物馆等主要机构）和后世声誉综合计算。
Images sourced from Wikimedia Commons, supplemented with a prestige score derived from institutional collection tier (Van Gogh Museum, Musée d'Orsay, MoMA, etc.) and posthumous critical reputation.
每封信被解析为三个心理维度，共同构成梵高某一时刻的内在坐标。
Each letter is parsed into three psychological dimensions, together forming a coordinate of Van Gogh's inner world at a given moment.
情绪的整体强度与平衡性。基于 valence（效价）、arousal（唤起度）和 dominance（支配感）三个 VAD 维度综合计算。
Overall emotional intensity and balance, derived from valence, arousal, and dominance (VAD) dimensions of affective language.
与他人关系的质量。基于 NRC 情感词汇中 trust（信任）、anger（愤怒）、fear（恐惧）等情感词汇的分布比例。
Quality of interpersonal relations, measured by the distribution of trust, anger, fear, and related emotion words from the NRC Emotion Lexicon.
创作生命力与自我拓展的程度。基于工作强度词汇、积极预期情感和不确定性表达的综合指标。
Creative vitality and self-expansion, based on work-intensity vocabulary, anticipatory positive affect, and expressions of uncertainty and exploration.
- 使用 NRC Word-Emotion Association Lexicon（Mohammad & Turney，2013）提取每封信的情感词频分布
- VAD 维度来自 NRC Valence, Arousal, and Dominance Lexicon，对信件词汇取加权均值
- 工作强度指标通过绘画/颜色/构图等领域词汇的出现频率计算
- 所有分数经过 z-score 标准化并映射至 [0, 1] 区间
- Emotion word frequencies extracted using the NRC Word-Emotion Association Lexicon (Mohammad & Turney, 2013)
- VAD dimensions from the NRC Valence, Arousal, and Dominance Lexicon; weighted mean over letter vocabulary
- Work-intensity indicator derived from frequency of domain words related to painting, color, and composition
- All scores z-score normalized and mapped to [0, 1]
这是一个语言模型，不是诊断工具。它描述的是信件中的语言特征，而非梵高本人的心理状态。原信以荷兰语和法语写成，本分析基于英译版，翻译过程中不可避免地有语义损耗。所有分数都应被视为近似的、探索性的。
This is a language model, not a diagnostic tool. It describes linguistic features of the letters, not Van Gogh's mental state per se. The original letters were written in Dutch and French; this analysis is based on English translations, which inevitably involve some semantic loss. All scores should be treated as approximate and exploratory.
| 时间轴Timeline | 时间轴分为三段：序章（1853–1880）压缩 12 倍，主线段（1880–1890）1:1 展开，尾声（1890）压缩 30 倍。这种非线性设计让创作期的密度得以完整呈现。 The timeline is divided into three segments: prologue (1853–1880) compressed 12×, main period (1880–1890) at 1:1 scale, and epilogue (1890) compressed 30×. This non-linear design preserves the density of his productive years. |
| 垂直位置Vertical pos. | 画作的垂直位置反映 Growth 维度——越靠上，生命力越强、创作越旺盛。信件的垂直位置则固定在中线附近，通过旋转角度传递关系信息。 Painting vertical position encodes the Growth dimension — higher placement indicates stronger creative vitality. Letters are anchored near the centerline; their rotational angle conveys relational quality. |
| 旋转角度Rotation | 信件的倾斜角度反映 Relation 维度：平衡倾向 0°，冲突倾向正负偏转。 Letter card rotation encodes the Relation dimension: harmonious letters lean near 0°, conflicted letters tilt more sharply. |
| 细节层次LOD | 缩放倍率较低时，邻近的元素自动聚合为色块和数量标注；放大后逐渐显示画作图像和信件卡片的完整细节。 At low zoom, nearby elements aggregate into color clusters with count labels. As zoom increases, individual painting images and letter card details progressively reveal. |
- 梵高博物馆（阿姆斯特丹）——梵高信件与画作研究的核心机构，本项目所用大量基础数据来源于此。 Van Gogh Museum, Amsterdam — the central institution for research on Van Gogh's letters and paintings, and the primary source of foundational data used in this project.
- Van Gogh Letters Project（vangoghletters.org）——Leo Jansen、Hans Luijten、Nienke Bakker 主编，提供了完整、严谨的英译全集。 Van Gogh Letters Project (vangoghletters.org) — edited by Leo Jansen, Hans Luijten, and Nienke Bakker, providing the complete and rigorous English translation used in this analysis.
- Saif M. Mohammad & Peter D. Turney，NRC Emotion Lexicon 的作者，为情感分析提供了关键的词汇资源。 Saif M. Mohammad & Peter D. Turney, authors of the NRC Emotion Lexicon, whose work underpins the sentiment analysis methodology.
梵高靠弟弟 Theo 的资助才能创作。如果这个项目对你有价值，
Van Gogh painted because his brother Theo believed in him. If this project means something to you,
请我喝杯咖啡 / Buy me a coffee ↗ Buy me a coffee ↗