Restoring History: Designing an African American Research Database
by Jennifer Gregory and Bridget B. Striker
Throughout Northern Kentucky (and particularly Boone County), there is a gap in knowledge regarding enslavement, the Underground Railroad (UGRR) movement, and African American history as a whole. In slaveholding border counties along the Ohio River, information about enslavement and UGRR activities—and African American voices in general—were suppressed for 150 years due to a complex societal situation and a mass exodus of former enslaved people in the years following the Civil War. According to the 1860 census, more than 20% of Boone County’s population was African American, and all but 45 of those individuals were enslaved.1 In contrast, by 1900, African Americans accounted for less than 5% of the county population.2 The African American population disappeared, and there was no community memory of why they left or where they migrated.
|Most importantly, the project gives voice to a population within Kentucky that was silenced for centuries.
A vast amount of information documenting the African American experience is interwoven into Kentucky records. In addition, well-known, documented UGRR routes passed through Northern Kentucky into Indiana and Ohio. However, Boone County had little evidence of UGRR activity—or any substantial discussion of slavery within the community beyond the story of Margaret Garner’s 1856 escape from enslavement in Richwood, Ky.3
For the past decade, Northern Kentucky has been moving toward a greater understanding and acknowledgment of local enslavement and the struggle for freedom. Boone County Public Library’s (BCPL) local history department recognized the need for a comprehensive initiative to compile and disseminate African American and UGRR history in Northern Kentucky to give voice to a population that was silenced for centuries.
The African Americans of Boone County Initiative was launched in 2014 to restore the historical African American voice in Boone County and to connect the descendants of those who were enslaved to that history, resulting in several projects. The Underground Railroad in Boone County project produced a bus tour, a website (railroad.bcplhistory.org), and a variety of classroom presentations and field trips. Additionally, in 2017, BCPL was recognized by the National Park Service as a Network to Freedom-affiliated member, both as a research facility and for our UGRR bus tour. Since then, BCPL has received two Network to Freedom-funded project grants. The various projects have been constructed like building blocks, with the intensive and systematic research needed at the local level for each project feeding into the next. While BCPL initially intended to focus entirely on Boone County, we discovered a trove of information regarding the entire Northern Kentucky region.
BCPL found that information about enslaved African Americans was hidden in plain sight among many primary historical resources, such as family papers, church rolls, and tax records, and we struggled to bring all of the information together to create one cohesive narrative. BCPL collected information on more than 3,300 enslaved people, their descendants, and the slaveholders. As the data steadily accumulated, BCPL struggled to keep the information organized. Thousands of records were stored in various files and formats on the shared BCPL local history department server. It was challenging for staffers to keep track of the multiple spreadsheets and other documents—and next to impossible to make the collected information accessible to the public.
The solution was found in a conference presentation4 by two librarians from Bowling Green State University (BGSU) in Ohio. They explained how they had used the Omeka S software platform as the basis for a linked data project, building a database of information pertaining to an annual campus music festival and then making it available to the public. BCPL now had a model for creating the African Americans of the Kentucky Borderlands database.
In 2019, BCPL received $50,000 in historic preservation mitigation funding for the project. The project consolidates the historical information into one cohesive platform, which allows the data to speak for itself and captures the essence of the lives of enslaved African Americans in Boone County. The end product is an online, fully searchable, integrated database of all extant information related to African American history and the UGRR within Boone County and the surrounding region.
Building a Database
When BCPL decided to use BGSU’s project as a model, we called the school for help. BGSU offered three key pieces of advice: Know the limitations of your software before you begin, severely limit the number of people with staff access to the database, and most importantly, stay focused on the data relationships.
BCPL chose to use the same software as BGSU, the open source Omeka S platform. We were told that the software had a functional limit of 200 (200 data links per item, 200 items per item set, etc.). We took this limit into account as we structured the data for the project. We learned how to use the software by experimenting with a set of test data. It quickly became obvious that even two staff members accessing the database was problematic. Keeping the data relationships in mind while determining how the data all linked together in the platform was a complex endeavor, making coordination among staffers next to impossible. In the end, the mitigation funds were used to hire a data specialist who uploaded and linked the data under our guidance.
The most important piece of advice we received from BGSU was to stay focused on the relationships between the individual pieces of data and not to get distracted by semantic web ontologies and their possibilities. As BCPL designed the structure for the data, we kept this in mind. Whenever obstacles were encountered, refocusing on the data relationships always cleared the path forward.
During this initial conceptual phase, the project team analyzed the data and data relationships, as well as considered ways in which researchers might choose to interact with the data. We eventually determined that the data naturally grouped into five categories: sources of information, locations, organizations, events, and people. After the analysis was complete, BCPL began the second phase of the process, the construction phase. The first step was to build the tools we would need. This involved choosing the semantic web ontologies we would use to describe the data and developing our templates for our five categories of information.
Once the tools were developed, BCPL contracted with a third party to host the Omeka S platform. The project team uploaded the ontologies we had selected and constructed the templates. At that point, the team began compiling the actual data in large spreadsheets. This was when standardization became important. We had to make some basic decisions about terminology. Was it the Berkshire family or the Barkshire family? What about the towns that no longer exist but have multiple historical names?
BCPL was fortunate in that we had already answered many of these questions. Three years before beginning this project, we endured a painful CMS migration, moving more than 10,000 local history digital assets from one platform to another, including migrating metadata from MARC to Dublin Core. To enable the metadata migration, BCPL built a local thesaurus of more than 2,000 terms, establishing the names we would use for towns, families, churches, and more. Over the course of constructing the African American database, we added approximately 500 more terms.
However, terminology wasn’t BCPL’s only issue. We were surprised to discover just how incomplete our information was. Seeing everything neatly arranged in a spreadsheet format made the gaps blindingly obvious. It was helpful to have additional staff available, and other members of the local history department contributed to filling in the information gaps.
Once each spreadsheet was as complete as possible, the project data specialist would upload it and interlink the datapoints. Omeka S has a feature that creates links during the uploading process, automatically linking to the data already in the database. Taking advantage of the feature meant carefully planning the order in which to upload the sheets. For example, a great deal of data may link to a newspaper as a source, but the newspaper itself doesn’t link to anything else. As a result, the least complicated data—sources—went in first, moving through the categories in order of complexity, finishing with the category requiring the most linking: people.
Once there was sufficient data loaded to work with, the project team began to build the public-facing website, our research portal. Omeka S has a website module built into the platform. BCPL’s webmaster added a banner and background and customized the website colors. The out-of-the-box website has several options for menus and other navigation, as well as both browsing and searching capabilities.
So, You Want to Do a Linked Data Project?
Building the database was different from many of the other digital projects BCPL has tackled in the past because of what we didn’t need. For example, we didn’t need many staffers. In fact, the fewer staff members involved, the better. The team also didn’t need a lot of IT expertise. BCPL used a third-party, cloud-based hosting service, and our webmaster created a minimal amount of customization for the public site, none of which was strictly necessary to operate. We also didn’t need a great deal of money, relatively speaking. BCPL was fortunate to have funding available that allowed us to hire a data specialist to upload the initial spreadsheets, but BCPL staffers could have done the work. It simply would have taken more time. The hosting service for Omeka S wasn’t very expensive, and the software platform itself was free.
All of this makes a linked data project like ours feasible for smaller organizations with fewer resources. The only thing the project truly required was two experienced librarians: one who was an expert in the data and another who was able to manage the metadata and impose standardization. Any experienced cataloger should be up to the task.
The only other requirements are the data and a platform to mount it on. It is helpful, but not vital, to have a thesaurus of local terms and a CMS. It is likely that if you don’t already have them when you begin, you will develop them in concert with your database.
The project website, African Americans of the Kentucky Borderlands, is available at omekas.bcplhistory.org/s/borderlands; data uploads are continuing. With roughly 6,000 entries in the initial spreadsheets for the enslavement era alone, the process takes time. Eventually, the database will encompass African American history for the region into the modern era. Work is now underway to prepare the data on the population migrations in the late 19th and early 20th centuries.
The African Americans of the Kentucky Borderlands database project is connecting researchers to the data, but it has also served to broaden the conversation about a missing chapter in the history of the region. Most importantly, the project gives voice to a population within Kentucky that was silenced for centuries. Through it, BCPL has helped bring awareness to the issues of enslavement within Northern Kentucky’s history and aided the region in acknowledging its past and those who were enslaved.