一、评价依据
A company's comprehensive value is hard to capture with a single indicator. This algorithm draws on multi-attribute utility theory (Keeney & Raiffa, 1976) to evaluate companies across four dimensions: Scale answers "how big," Quality answers "how strong," Innovation answers "how new," and Contribution answers "how substantial." Among them, the Contribution dimension is a unique CPPPS feature—taxes reflect returns to public finance, employment reflects the carrying of social livelihoods, and real-economy attributes reflect support for material production and industrial chains—something most capital-return-oriented international business rankings overlook.
企业综合价值难以用单一指标刻画。本算法借鉴多属性效用理论（Keeney & Raiffa，1976），从四个维度立体评价：规模回答"有多大"，质量回答"有多强"，创新回答"有多新"，贡献回答"有多实"。其中贡献维度是 CPPPS 的独有特色——税收体现对公共财政的回馈，就业体现对社会民生的承载，实体经济属性体现对物质生产与产业链的支撑，这是多数以资本回报为导向的国际商业榜单所忽略的。
- 规模维度：营收 + 总资产，衡量企业经营的体量与资源掌控力。
- 质量维度：净利润 + 市值，衡量盈利质量与资本市场认可度。
- 创新维度：研发投入 + 科技属性，衡量技术密集度与未来成长性。
- 贡献维度（CPPPS 特色）：税收 + 就业 + 实体经济属性，衡量企业对财政、民生与实体经济的回馈。
二、数学模型
Each indicator first undergoes normalization and logarithmic transformation, is then weighted into its dimension, and is finally aggregated by weighting into a comprehensive index:
各指标先经归一化与对数变换，再按维度加权合成，最后加权汇总为综合指数：
The weights of the four dimensions always sum to 1, and are all public, reproducible, and auditable.
四个维度权重之和恒等于 1，且全部公开、可复现、可审计。
三、指标与权重依据
| 维度 | 指标 | 方向 | 维内权重 | 维度权重 | 依据 |
|---|
| 规模 | 营收 | 正向(对数) | 0.60 | 0.30 | 经营体量核心 |
| 规模 | 总资产 | 正向(对数) | 0.40 | 资源掌控力 |
| 质量 | 净利润 | 正向(对数) | 0.50 | 0.25 | 盈利质量 |
| 质量 | 市值 | 正向(对数) | 0.50 | 市场认可 |
| 创新 | 研发投入 | 正向(对数) | 0.60 | 0.20 | 技术投入强度 |
| 创新 | 科技属性 | 正向 | 0.40 | 技术密集度 |
| 贡献 | 税收 | 正向(对数) | 0.35 | 0.25 | 财政贡献 |
| 贡献 | 员工人数 | 正向(对数) | 0.35 | 就业贡献 |
| 贡献 | 实体经济属性 | 正向 | 0.30 | 实体/产业链支撑 |
四、数据归一化与对数变换
Different indicators differ greatly in dimension (revenue, assets, and market value are measured in hundreds of billions of dollars, while employees are measured in tens of thousands), so dimensions must first be removed. For attribute-type indicators (technology attributes, real-economy attributes), this algorithm uses linear min-max normalization:
不同指标量纲差异巨大（营收、资产、市值以千亿美元计，员工以万计），须先消除量纲。本算法对属性类指标（科技属性、实体经济属性）采用线性 min-max 归一化：
For monetary and headcount indicators with extremely wide ranges (revenue, assets, profit, market value, R&D, taxes, employees), the logarithm is taken before normalization to prevent a few leading companies from monopolizing the scale and to keep the ranking more robust:
对规模跨度极大的金额类与人数类指标（营收、资产、利润、市值、研发、税收、员工），先取对数再归一化，避免头部少数企业垄断、拉开与中后部差距，使排名更稳健：
Logarithmic transformation reflects "diminishing marginal returns": an increase in revenue from $100 billion to $200 billion is less meaningful than an increase from $10 billion to $20 billion. When all values of an indicator are identical, a uniform value of 0.5 is used to avoid division by zero.
对数变换体现"边际递减"：营收从 1000 亿美元增至 2000 亿的增量意义，小于从 100 亿增至 200 亿的增量意义。当某指标全体相同时，统一取 0.5 以避免除零。
五、数据来源与局限性
The revenue, profit, assets, market value, R&D, taxes, and employee data in this ranking all use publicly known approximations, sourced from companies' public annual reports, financial information platforms, and industry statistics, marked "subject to official data"; technology attributes and real-economy attributes are qualitative scores based on companies' main businesses. Differences exist across countries in accounting standards, fiscal-year conventions, and exchange-rate conversion, and approximations may deviate from official exact values; therefore, this ranking is for research and reference only and does not constitute any investment advice.
本排名的营收、利润、资产、市值、研发、税收、员工数据均采用公开常识近似值，来源为各企业公开年报、财经资讯平台及行业统计的公开数据，标注"以官方为准"；科技属性与实体经济属性为基于企业主营业务的定性评分。不同国家会计准则、财年口径、汇率折算存在差异，近似值与官方精确值存在误差，故本排名仅供研究与参考，不构成任何投资建议。
六、参考文献
- Keeney, R. L., & Raiffa, H.（1976）：Decisions with Multiple Objectives: Preferences and Value Tradeoffs，多属性效用理论。
- Forbes：Global 2000 榜单方法论（以营收、利润、资产、市值四项综合的公开参照框架）。
- 《财富》世界500强榜单（以营业收入排序的国际参照）。
- 各企业年度报告、可持续发展（ESG）报告等公开披露数据。
- 全球主要证券交易所公开市值数据（财经资讯平台汇总）。