EAGER: Ecosystem Modeling for Infrastructure Evaluation

本研究调查了科学软件开发人员、维护者、用户和资助者如何以最大化科学软件开发资金的有效性为目标进行交互。这些群体之间的相互作用通过研究生态系统的技术进行建模和分析。生态系统的类比被成功地用于描述相互作用的实体之间的关系，特别是当组织可能在某些活动上合作而在其他活动中竞争时。这些技术支持对大量工件的表示和分析，例如集成成一个完整软件包的数千个单独文件。这些关系是多种多样的，因此使用基本的图表示来对交互进行建模。然后，使用各种算法计算图的属性，如中间性中心性，从中可以导出关系的特征。这些信息可用于确定软件的某些质量、使用的多样性，以及从其他项目和库中重用的代码数量。这项研究将对社会技术生态系统方法做出根本性的贡献，其形式是制定生态系统可行性决策的新技术。这项研究的结果将帮助开发经理和资金经理通过解决长期可行性和可持续性问题做出有效的决策。这包括在科学基础设施中使用的特定软件模块和软件包的适当性。将开发用于评估软件的新技术，以供组织采用、使用或包含在其自己的产品中。通过采用全面的生态系统方法和使用基于图形的质量度量，组织将能够更准确地估计最初更新和最终维护软件组件所需的工作量。这些结果将展示开发的技术如何能够帮助管理科学项目中的风险，以及这些技术将如何导致更有效地利用资源，包括联邦资金。

This research investigates how scientific software developers, maintainers, users, and funders interact, with the goal of maximizing the effectiveness of funds for scientific software development. The interactions among these groups are modeled and analyzed through techniques used to study ecosystems. The ecosystem analogy is being used successfully to describe the relationships among interacting entities particularly when organizations may collaborate on some activities and compete in others. These techniques support the representation and analysis of large numbers of artifacts such as the thousands of individual files integrated to form a complete software package. The relationships are diverse, so a basic graph representation is used to model the interactions. Then, a variety of algorithms are used to compute graph attributes such as betweenness centrality, from which characteristics of the relationships can be derived. This information can be used to determine certain qualities of the software, the diversity of its use, and the amount of code reused from other projects and libraries. This research will make fundamental contributions to socio-technical ecosystem methods in the form of new techniques for making decisions about the viability of an ecosystem. The results of this research will help development managers and funding managers make effective decisions by addressing long-term viability and sustainment issues. This includes the appropriateness of specific software modules and packages for use in scientific infrastructures. New techniques for evaluating software for adoption by an organization for use or inclusion in their own products will be developed. By taking the comprehensive ecosystem approach and using graph-based quality metrics, an organization will be able to more accurately estimate the effort that will be required to initially renovate, and eventually maintain, a software component. The results will demonstrate how the developed techniques can help to manage risk in a science project, and how these techniques will result in more efficient use of resources, including federal funding.