Abstract Free and open source software package ecosystems have existed for a long time and are among the most sophisticated human-made systems. One of the oldest and most popular software package ecosystems is CRAN, the repository of packages of the statistical language R, which is also one of the most popular environments for statistical computing nowadays. CRAN stores a large number of packages that are updated regularly and depend on a number of other packages in a complex graph of relations; such graph is empirically studied from the perspective of complex network analysis (CNA) in the current article, showing how network theory and measures proposed by previous work can help profiling the ecosystem and detecting strengths, good practices and potential risks in three perspectives: macroscopic properties of the ecosystem (structure and complexity of the network), microscopic properties of individual packages (represented as nodes), and modular properties (community detection). Results show how complex network analysis tools can be used to assess a package ecosystem and, in particular, that of CRAN.

中文翻译：

---

综合 R 档案网络 (CRAN) 包生态系统的复杂网络分析

摘要 免费开源软件包生态系统已经存在了很长时间，是最复杂的人造系统之一。最古老和最受欢迎的软件包生态系统之一是 CRAN，它是统计语言 R 的软件包存储库，也是当今最流行的统计计算环境之一。CRAN 存储了大量定期更新的包，并依赖于复杂关系图中的许多其他包；本文从复杂网络分析 (CNA) 的角度对此类图进行了实证研究，展示了先前工作提出的网络理论和措施如何帮助分析生态系统并从三个角度检测优势、良好做法和潜在风险：生态系统的宏观属性（网络的结构和复杂性）、单个包的微观属性（表示为节点）和模块化属性（社区检测）。结果显示了如何使用复杂的网络分析工具来评估软件包生态系统，尤其是 CRAN 的生态系统。