Relationships between legal entities can be represented as a large weighted directed graph. In this work, we model the global capital ownership network across a hundred of millions of such entities with the goal of establishing a methodology for extracting and analysing meaningful patterns of capitalistic influence from the graph structure. To do so, we adapt and employ metrics from graph analytics and algorithms from the area of influence maximization. We characterize the relationships extracted and show that our analysis aligns with information from macro-economic studies; for example it recovers the presence of known tax heavens, which appear in dense subgraphs of countries. We also identify and quantify cases where capital is principally owned by others, corresponding to global influence. Beyond confirming known patterns and justifying our novel application of influence maximization methodology in this area, the outcome also offers new insight and metrics in this domain, by highlighting the existence of strong communities of capitalistic property. We leverage influence maximization methods as a means to evaluate the impact of entities in these contexts. Finally we formulate the results of our study into recommendations for future analyses of this kind.

法人实体之间的关系可以表示为一个大型加权有向图。在这项工作中，我们对亿万个这样的实体的全球资本所有权网络进行建模，目的是建立一种从图结构中提取和分析有意义的资本主义影响模式的方法。为此，我们根据影响力最大化的领域来调整和应用来自图分析的度量和算法。我们表征了提取的关系，并表明我们的分析与宏观经济研究的信息相符。例如，它恢复了已知税收天堂的存在，这些税收天堂出现在国家的密集子图中。我们还确定并量化了与全球影响力相对应的资本主要由他人拥有的情况。除了确认已知的模式并证明我们在这一领域中最大程度地运用影响力最大化方法论的合理性之外，其结果还通过突出了强大的资本主义财产共同体的存在，为这一领域提供了新的见识和衡量标准。我们利用影响最大化方法来评估这些实体在这些情况下的影响。最后，我们将研究结果整理成建议，以供将来进行此类分析。