Evolution of an application profile: advancing metadata best practices through the Dryad data repository

Dryad is a general-purpose curated repository for data underlying scholarly publications. Dryad’s metadata framework is supported by a Dublin Core Application Profile (DCAP, hereafter referred to as application profile). This paper examines the evolution of Dryad’s application profile, which has been revised over time, in an operational system, serving day-to-day needs of stakeholders. We model the relationships between data packages and data files over time, from its initial implementation in 2007 to its current practice, version 3.2, and present a crosswalk analysis. Results covering versions 1.0 to 3.0 show an increase in the number of metadata elements used to describe Dryad’s data objects in Dryad. Results also confirm that Version 3.0, which envisioned separate metadata element sets for data package, data files, and publication metadata, was never fully realized due to constraints in Dryad system architecture. Version 3.1 subsequently reduced the number of metadata elements captured by recombining the publication and data package element sets. This paper documents a real world application profile implemented in an operational system, noting practical system and infrastructure constraints. Finally, the analysis presented informs an ongoing effort to update the application profile to support Dryad's diverse and expanding community of stakeholders.

达德里亚德仓储（Dryad）是一款面向学术出版物配套数据的通用编目仓储库。其元数据框架依托都柏林核心应用框架（Dublin Core Application Profile，缩写为DCAP，下文简称应用框架）。本文探讨了达德里亚德仓储应用框架在实际运行系统中的演进历程——该框架历经多次修订，以满足利益相关方的日常业务需求。我们以时间为轴线，从该框架2007年首次落地至今的3.2版本，对数据包与数据文件间的关联关系进行建模，并开展了交叉映射分析。针对1.0至3.0版本的分析结果显示，达德里亚德仓储中用于描述数据对象的元数据元素数量呈增长趋势。分析结果同时证实：3.0版本曾计划为数据包、数据文件及出版物元数据分别设置独立的元数据元素集，但受限于达德里亚德仓储的系统架构，该设计从未完全落地。随后推出的3.1版本通过合并出版物元数据与数据包元数据元素集，缩减了所需采集的元数据元素总量。本文记录了一套在实际运行系统中落地的真实应用框架，并着重指出了其面临的实际系统与基础设施限制。最后，本次分析可为当前正在推进的应用框架更新工作提供参考，以更好地服务达德里亚德仓储日益多元化且规模不断扩大的利益相关方群体。