This report shares ongoing metadata remediation work on The University of Michigan Library’s Deep Blue Repositories, or the “Deep Blues,” primarily focused on Deep Blue Documents. Deep Blue Documents is a repository that has been accepting research outputs since 2006. Currently, it contains about 160,000 deposits of items from 1863 to the present, with hundreds of thousands of downloads per month. Deep Blue Data launched in 2016, and as of Fall 2025 has approximately 2,000 deposits. Deep Blue Documents is structured as a series of parallel collections, based in many cases on departmental, institute, center, or school structures (such as the University of Michigan Biological Station collection), and in others on particular ongoing projects (such as the Global Feminisms Project). A small number of collections, including the Dissertations and Theses collection, represent materials spanning the University’s entire disciplinary coverage and timeline–in this case, beginning with the early 20th century. The Deep Blues are housed within the University of Michigan Library system, where there are ongoing efforts for metadata repair and remediation. Many similar institutions are undertaking reparative description work as well, and there is a growing body of literature on these topics. However, there is still a need for best practices and applicable general principles to guide this work, especially for self-deposit repositories. We hope this report will add to the growing body of documentation on how to put theory into practice within the constraints of the limited resources that many repositories face. This report begins with a brief description of how we conceptualize reparative description, sharing the five guiding principles we see as critically important when engaging in this work. These are 1) Embracing intentional, flexible practices that allow for changing course; 2) Building plurality or multiple points of view within metadata; 3) Unlearning the “neutral” voice of traditional description; 4) Prioritizing collaborative relationships with community stakeholders; and 5) Documenting work practices, ultimately contributing to making them visible and reproducible for future practitioners. We next move into describing work practices commonly seen as integral to reparative description, including contextual statements and metadata repair. This section also speaks to these practices' general applicability to Deep Blue Documents. Next, we share case studies drawn from three collections within Deep Blue Documents that provide unique contexts for metadata remediation: 1) the UM Biological Station collection, 2) the Global Feminisms Project collection, and 3) the Dissertations and Theses collection. These case studies include a summary of each collection, preliminary findings from consulting with collection owners, and recommendations for possible future remediation directions and practices. Finally, since reparative work is a process, not an end product, we conclude with suggestions for how to sustain these considerations in future deposits, and overall remediation considerations for the repository moving forward. In general, this document recommends a nuanced approach to metadata remediation based on the needs and context of the collections within the institutional repository, not blanket changes of metadata to remove or censor harmful content. We hope this document is useful to other collections or divisions at the University of Michigan who are interested in reparative metadata work, but don’t have a clear starting point. Further, we hope this document will engage other institutional repositories who are similarly interested in this work but are also struggling with how to remediate and manage metadata that is depositor-created and thus intentionally not controlled. For any feedback or questions about the work described in this report, please contact deepblue at umich.edu. You can also find this reports’ recommendations summarized in its Appendix B: Reparative Description and Metadata Remediation Recommendations Summary.
Rachel et al. (Mon,) studied this question.