SI2-SSI: CRESCAT, A Computational Research Ecosystem for Scientific Collaboration on Ancient Topics, Spanning the Full Data Life Cycle

该项目集成、测试并记录了一套可互操作的软件工具，以支持协作研究。这些工具统称为CRESTAT(古代主题科学合作的计算研究生态系统)。最初的重点是研究过去很长一段时间内处于空间位置的种群内的动态相互作用和结构变化的学科，例如古生物学、考古学和经济史。尽管它们不同，但这些学科在建模和分析数据方面有相似的计算需求。此外，同样的软件可用于许多其他学科，通过建立和维护一套通用的可互操作工具来服务于广泛的研究人员，从而实现规模经济，同时跨越研究数据的整个生命周期，包括(1)获取、(2)整合、(3)分析、(4)出版和(5)数据存档。为终端用户研究人员提供了直观的图形用户界面，使他们可以在生命周期的所有阶段处理数据，而无需繁琐的手动数据传输和转换。该项目将解决一个影响许多科学学科的主要计算问题，因为基于不同的空间、时间和分类本体集成和分析不同来源的数据是一项挑战。因此，它将通过展示如何明确地表示个人判断的全部可变性以及表达这些判断的不同的概念化和术语，并明确地将每个观察、解释和概念本体论归因于特定的被命名的个人或群体，在科学界和其他领域产生广泛的影响。与许多科学研究的计算工具不同，CRESTAT假设了不存在的一定程度的本体论共识，它符合实际的研究实践。它没有强加一个标准化的本体论，从而忽视或压制了研究人员之间不可避免的分歧和相互冲突的解释。相反，它代表了一个更大的公共框架内的本体论多样性、观察不确定性和解释分歧，最终用户可以在其中查询、分析和比较所有的观察、解释和术语，以告知他们自己对证据的判断。CRESTAT旨在允许以数字方式表示科学分歧以及观察和解释的不确定性，从而将这些差异本身暴露为分析和辩论的数据。因此，除了为高级研究建立一个更有效的共享框架的实际目标之外，拟议的工作还将引发关于计算工具应该如何与科学实践相关的理论思考。CRESTAT项目是计算机科学家、古生物学家、地球科学家、考古学家、经济历史学家和其他社会科学家之间的跨学科合作。其目标是展示跨越社会科学和自然科学的综合软件生态系统的价值，并可促进任何以时间和空间关系模型重叠或术语和分类相互冲突为特征的研究。克雷斯卡特对科学知识的表述避免了强制标准化，这在许多情况下是不切实际的，因为缺乏执行机制，而且在原则上也是有问题的，因为不同的本体论往往合法地反映了不同的理论假设和研究议程。CRESTAT工具套件的核心是一个创新的数据集成系统，它明确地表示研究数据和数据中固有的本体。本体论在这里被定义为给定知识领域中的实体及其之间关系的概念模型，与模式相反，模式是在工作系统内的逻辑数据结构中实现本体论。克雷斯卡特的数据集成系统在抽象水平上运行，足以提供基于抽象全球模式的可预测和可高效查询的数据库结构，而抽象全球模式又基于适用于所有科学和学术学科的基本概念和关系所规定的上层本体论。数据集成系统是在企业级XML/XQuery DBMS中实现的，该数据库充当数据仓库(使用非关系图形数据模型)，其中存储了来自代表许多学科的各种研究项目的各种数据。每个研究项目的术语和概念区别都得到了完全保留。在CRESTAT项目中采用的研究数据的方法是：(1)在单一分析框架内连贯、紧密地集成软件工具和数据格式；(2)开放式、互连现有工具，同时允许在未来添加新工具；(3)非排他性，绝不阻止其组件工具参与其他软件生态系统；(4)可伸缩，旨在处理大规模数据管理、分析和可视化；以及(5)可持续，维护共享资源，以满足软件和技术支持的共同需求，从而实现显著的规模经济。

This project integrates, tests, and documents a suite of interoperable software tools to support collaborative research. The tools are collectively called CRESCAT (Computational Research Ecosystem for Scientific Collaboration on Ancient Topics). The initial focus is on disciplines that deal with dynamic interactions and structural changes within spatially situated populations over long time spans in the past, e.g., paleobiology, archaeology, and economic history. Despite their differences, these disciplines have similar computational needs for modeling and analyzing data. Moreover, the same software can be used in many other disciplines, enabling economies of scale by building and maintaining a common set of interoperable tools to serve a wide range of researchers, while spanning the full research data life cycle, consisting of (1) acquisition, (2) integration, (3) analysis, (4) publication, and (5) archiving of data. An intuitive graphical user interface is provided for end-user researchers to work with their data in all stages of the life cycle without cumbersome manual data transfers and transformations. The project will address a major computational problem that affects many scientific disciplines due to the challenge of integrating and analyzing data of diverse origins based on heterogeneous spatial, temporal, and taxonomic ontologies. Thus it will have a broad impact in the sciences and beyond by showing how to represent explicitly the full variability of individual judgments and the divergent conceptualizations and terminologies through which those judgments are expressed, with explicit attribution of each observation, interpretation, and conceptual ontology to a particular named person or group. Unlike many computational tools for scientific research, which assume a degree of ontological consensus that does not exist, CRESCAT conforms to actual research practices. It does not impose a standardized ontology, thereby ignoring or suppressing the inevitable disagreements and conflicting interpretations that arise among researchers. Instead, it represents ontological diversity, observational uncertainty, and interpretive disagreement explicitly within a larger common framework in which end users can query, analyze, and compare the full range of observations, interpretations, and terminologies to inform their own judgments about the evidence. CRESCAT is designed to allow scientific disagreements and observational and interpretive uncertainties to be represented digitally in a way that exposes these differences themselves as data for analysis and debate. Thus, in addition to the practical goal of building a more efficient shared framework for advanced research, the proposed work will provoke theoretical reflection about how computational tools should relate to scientific practice.The CRESCAT project is an interdisciplinary collaboration between computer scientists, paleobiologists, geoscientists, archaeologists, economic historians, and other social scientists. The goal is to demonstrate the value of an integrative software ecosystem that spans the social and natural sciences and can facilitate any research characterized by overlapping models of temporal and spatial relations or by conflicting terminologies and taxonomies. CRESCAT's representation of scientific knowledge eschews forced standardization, which is impractical in many cases due to lack of an enforcement mechanism and is also questionable in principle since divergent ontologies often legitimately reflect different theoretical assumptions and research agendas. Central to the CRESCAT suite of tools is an innovative data-integration system that represents explicitly both research data and the ontologies inherent in the data. An ontology is defined here as a conceptual model of entities and the relationships among them in a given domain of knowledge, in contrast to a schema,”which is the implementation of an ontology in logical data structures within a working system. CRESCAT's data-integration system operates at a level of abstraction sufficient to provide a predictable and efficiently queryable database structure based on an abstract global schema, which in turn is based on an upper ontology specified in terms of fundamental concepts and relationships applicable to all scientific and scholarly disciplines. The data-integration system is implemented in an enterprise-class XML/XQuery DBMS that serves as a data warehouse (using the non-relational graph data model), in which is stored diverse data from a wide range of research projects representing many disciplines. The terminology and conceptual distinctions of each research project are fully preserved. The approach to research data taken in the CRESCAT project is (1) coherent, tightly integrating software tools and data formats within a single analytical framework; (2) open-ended, interconnecting existing tools while allowing the addition of new tools in the future; (3) non-exclusive, in no way preventing its component tools from participating in other software ecosystems; (4) scalable, designed to handle large-scale data management, analysis, and visualization; and (5) sustainable, maintaining shared resources to meet common needs for software and technical support and thus enabling substantial economies of scale.