Ontology-based Data Organization for the Enslaved.org Project

The men, women, and children forced into slavery in the Atlantic world came from diverse African societies with long histories of political, economic, and cultural development. They were taken from the trading centers of the Hausa city-states, the farming and artisanal communities of Senegambia, the Kongo and Mbundu polities of West Central Africa, and many other regions. They carried with them agricultural expertise, metallurgical skills, medical knowledge, religious traditions, and oral histories that helped sustain communities in the face of displacement and enslavement.Enslavement did not erase this intellectual and cultural inheritance, nor did it render its victims passive numbers in history. Instead, these individuals—scattered against their will across the Americas—were bearers of cultures. In the markets of Havana, the rice fields of the Carolina Low Country, the sugar mills of Bahia, and the mountain estates of Jamaica, enslaved Africans repurposed inherited knowledge to meet new realities. They wove fragments of home into new forms of resistance, ritual, kinship, and survival. The historical record that bears witness to these lives is scattered—geographically dispersed, archivally fragmented, and digitally siloed. The traces of individuals often exist as mere shadows: a name in a baptismal registry, a notation in a ship manifest, a mention in a planter’s ledger, a deposition in a colonial court. These documents, while scattered across continents and disciplines, gesture toward one another—but too often speak in isolation.Enslaved: Peoples of the Historical Slave Trade confronts this problem directly. Through an ontology-based approach to data organization, it seeks not only to systematize knowledge of the historical records of enslavement, but to restore connective tissue between dismembered lives and histories. This is more than a technical endeavor; it is an ethical one. Each name, record, fragment is treated not as a data point but as the trace of a full human life—a person whose agency was never wholly extinguished. The ontology functions as both a scaffolding for the integration of disparate archival materials and as a principled commitment to historical justice. By bringing scattered records into structured relation, we begin the work of reassembling stories often left out of histories.1Imagine trying to trace a single enslaved person's life story using scattered digital archives. You might find 'Maria Angola' baptized in a Rio de Janeiro church record, 'Maria' listed as a seamstress in a plantation inventory from Bahia, and 'Maria Angola' mentioned in a manumission document from São Paulo. Are these the same person? Traditional databases cannot tell you; each institution stores its records separately, using different naming conventions, different categories, even different spellings of the same name. Linked data works differently. Instead of trapping information in isolated databases, it creates connections across collections, allowing 'Maria Angola' in Rio to link to 'Maria' in Bahia if the evidence supports it.Think of it as building bridges between islands of information, enabling researchers to follow individual lives across the fragmentary historical record in ways that were previously impossible. Yet bridges need blueprints. When one archive calls someone a 'seamstress' and another calls the same person a 'costureira,' how does the computer know they mean the same thing? When one database records 'baptism' and another 'batismo,' how do we connect these related events? An ontology serves as a shared vocabulary—a translator that helps different databases understand each other. It defines not just what terms mean, but how they relate to one another: that baptisms are religious ceremonies, that seamstresses perform skilled labor, that a person can have multiple names across different documents. Without this common language, linked data becomes a tower of Babel, technically connected but semantically meaningless. The ontology ensures that when we say 'Maria was baptized,' every database in the network understands both what baptism means and how it relates to Maria's larger life story. We must be careful to note that linking does not draw conclusions, but the ability to bring data together from disparate sources, allowing researchers to draw conclusions and make connections.In addition, the modular approach to ontologies recognizes a fundamental truth about historical research: different researchers need different levels of detail for different purposes.2 A genealogist tracing family connections cares deeply about kinship relationships but may not need elaborate information on economic transactions. A historian studying labor systems needs detailed occupational categories but might simplify location data. A modular ontology works like a toolbox of possibilities: researchers can select the components they need while ensuring everything still fits together. If someone later wants to add more nuanced categories for African ethnic groups or more precise models of ship voyages, they can build new modules that plug into the existing framework without breaking what is already there. This flexibility proves crucial for a project like Enslaved.org, spanning centuries, continents, and scholarly disciplines, allowing the ontology to grow and adapt as new research questions emerge while maintaining the connections that make the whole system valuable.In using linked data with a modular ontology, Enslaved.org has created a new model for collaborative humanities scholarship focusing on people of the historical slave trade. This model not only brought together several historians and a host of digital history projects and archives focusing on the historical slave trade, but it also required computer scientists like us who work on the semantic web. For us, the technical goal of Enslaved.org was to establish what the project calls the Enslaved.org hub — a one-stop querying and inspection tool for integrated historic data on the slave trade, originating from heterogeneous contributors and their data sources, thereby allowing students, researchers and the public to understand and reconstruct the lives of individuals who were part of the historical slave trade, most importantly enslaved peoples. As our part, to address the underlying data integration issues, we opted to construct a knowledge graph, expressed in RDF, with an underlying schema in the form of an OWL ontology. In this article, with a focus on the development of the ontology, we will explain this approach, and in doing so hope to make it more understandable for a general audience. Thus, we begin by defining the key concepts behind modular ontologies. Then we discuss the methodology used to construct the modular ontology for the Enslaved.org project. We then present examples from the Enslaved.org ontology and its documentation, respectively. Finally, we briefly conclude and also invite you to learn more about the behind the scenes work in the documentation section of Enslaved.org.Ontologies emerged in the 1990s as formal frameworks for organizing knowledge. They have been described as 'explicit shared specifications of conceptualizations,' and as such they indeed seem like a natural fit for such a role,3 with respect to interoperability4 and reusability of field specific knowledge and data. In essence, ontologies create a common vocabulary and structure that helps different people—and different computer systems—interpret information in the same way. They should be considered sophisticated classification systems that not only categorize items but also define how these categories relate to each other.5The research field known as 'Semantic Web' within Computer Science and adjacent disciplines has driven the development of corresponding standards.6 Proposed by World Wide Web inventor Tim Berners-Lee around 2001, the Semantic Web represented an ambitious vision to transform the internet from a collection of human-readable documents into a network of machine-interpretable data. This evolution promised to enable computers to process (understand) better the meaning of information rather than simply displaying it, potentially changing if not revolutionizing how we search for and use online resources.To realize this vision, researchers developed specialized languages and frameworks, both for ontologies (the Web Ontology Language, OWL7) and for actual content data, the Resource Description Framework (RDF8). OWL provides the sophisticated vocabulary needed to create detailed ontologies, whereas RDF offers a standardized way to express data as simple statements in a subject-predicate-object format, similar to basic sentences in human language. These technologies, although powerful, initially remained primarily in academic and specialized technical communities. Having standards and frameworks is key since it helps make all the information machine (e.g., computer) readable.In a newer development around 2006, the term 'Linked Data' has been used to describe shared data and accompanying metadata. Championed by Berners-Lee, Linked Data represented a more practical approach to implementing Semantic Web principles. The idea of linked data, structured and standardized machine-readable data, established a direct method for publishing data online that could connect related information across different sources. The implementation of linked data helped to bridge the gap between theoretical semantic models and the real-world applications.Popularized by Google's 2012 announcement of its Knowledge Graph, this concept describes vast networks of entities (people, places, concepts) and their interrelationships. A knowledge graph represents information as a web of interconnected facts, where individual pieces of data—a person's name, a ship's departure date, a plantation's location—exist not in isolation but as nodes linked by meaningful relationships that mirror how knowledge works in the real world. Unlike traditional databases that store information in rigid tables and columns, knowledge graphs capture the fluid, contextual nature of historical evidence. This structure proves particularly powerful for historical research because it mirrors how historians think—not in discrete categories but in networks of people, places, events, and relationships that intersect and influence each other across time and space.The graph becomes both a repository of facts and a discovery mechanism, allowing researchers to pose questions that traverse multiple archives and databases—research questions like 'trace all records mentioning someone named Maria Angola across different cities and decades'—and receive answers that emerge from the relationships encoded in the graph's structure, revealing connections that would remain invisible in traditional isolated databases. While building on Semantic Web foundations, knowledge graphs emphasize practical utility and have gained widespread adoption in commercial applications like search engines, virtual assistants, and recommendation systems.These technologies, together with common shared ontologies (e.g., Schema.org, Dublin Core, Vocabulary for Interlinked Datasets) allow for indices to be built, aiding in discovery. The use of established standards generally makes access and reuse easier due to robust and shared tooling infrastructure. This standardization has helped to create connections or bridges between different sets of data, enabling researchers to bring together data from multiple sources, and thus allowing for the discovery of connections that might otherwise remain hidden.Despite all these advances which already simplify data discovery and reuse, due to the sheer amounts of data generated every day, the management of such data is still a very significant drain on resources for both the data provider and the data user, and thus corresponding research continues to investigate to improve methods and tools for data of this research to the of how ontologies the or should be developed and such that data reuse, and which we in the Enslaved.org on a fundamental modular ontologies knowledge and breaking knowledge into interconnected but we have created a framework that can for their specific research This modular approach researchers to adapt of the ontology to their while crucial connections to the structure or even to developed by of it as a sophisticated researchers can or without the ability to connect with other work using the same This between and creates an where specialized research can without isolated information their most are simply an a of or a of structure, which is as In the of ontology we have an ontology These are that across different data with has established as an ontology that continues to be the of an is a a schema for an A schema is a of how concepts are related to each other It should be that schema to be an of the encoded knowledge. The are the The of a in the is used by us to where a concept and how it should be For the is the of the The other with that they are to the the that they are of its — it may be more than the or may be used to that is This is because are to be general as to to as many different as that the key is to how to create data models that address specific needs of a but general to be shared and schema of this that an has a it in and and that it provides for is or in the can be of other this is a very simple and it is to with which do not fit this is more to an than and The idea of a modular is not to and capture all of our but rather to simply model a for data that will be in many — not all or even most — use the Enslaved.org as crucial organizing that mirror their in historical This from interconnected the of information is in an a specific and a specific in historical records about enslaved individuals were most often created significant by the in This approach that historical documentation does not simply capture a of facts, but in time and each with its and that into what was and Enslaved.org Ontology is not only as of are but it is also as a that records are created primarily by the and colonial information around events, the ontology the nature of the historical record, a framework that more represents how knowledge about was and In this we to make our models as as with the most used data by the in a a general such as that in is too and for a use Thus, it is to create a corresponding from it by the to the specific and we have to the it was to the and of the into relationships of this is the of this The in in this a to a vocabulary of the ontology and can be on the Enslaved.org a formal of implementation of as modular components what we have a These modular ontologies their and Ontology traditional knowledge models that as modular ontologies the of their development and that documents not just what concepts exist but how and they were structured in specific This approach functions like that a revealing the underlying and that might otherwise remain invisible to to or the When researchers new data concepts as previously categories of labor or more nuanced of kinship or need to existing or these practical for which modular components can be which must be and how new can with the existing than an web of relationships and that can the modular ensuring that rather than while maintaining with related knowledge This is a key to the Enslaved.org Ontology and its new on the work a modular ontology of ontology This methodology functions to how building to knowledge while maintaining across diverse approach to a knowledge with on reusability and These enable easier integration with both and while also to knowledge in a that can be for multiple our ontology can adapt to diverse research questions while maintaining underlying theoretical and practical implementation of this modular ontology approach have been in and work Ontology which development and to knowledge the we a of to construct the Enslaved.org This structured approach that our knowledge framework both and technical its development we both the as a general process that can be for other projects and and examples from our work on the Enslaved.org in a modular ontology is to define the and use which we did with our on the Enslaved.org questions were in a of ways search were from from historians in slavery and The for Enslaved.org and how the different would have their and search on categories such as events, and and time project questions by about the data in natural language. we a of the historians thus a on the questions and the of data sources. the data to be into the Enslaved.org were African The Slave Trade Slave of and of African and African of and Slave as they a of different of data and as a of key was a where data and ontology the modules and the ontology. to the focus on historical data, the categories of and The content focus of the ontology is on and key data, such as their name, (e.g., Enslaved.org or or or not the to they in events, and relationships to other or such as family or As the and records of a in the data, as historic records from specific (e.g., allow this data to connect to it was to model a way for of data, and specific research projects ontology as a for corresponding were such as using the of for using of person in to when for in and for the other of the data, such as and to be represented with in to more the data, as as An was as the data it both and in to the of to of data, and opted to simply a name that a simple to be more a later a of historic was out of for the of the and so we a approach with such as historical the focus is on historical data, as by sources, it was also that the and the knowledge graph, would data, rather than a which required careful in our approach to the knowledge graph schema this we the nature of by to historic data about in the form of rather than about individual the of the concept of the of and we this to section where we briefly describe of the modules of the Enslaved.org the ontology and can be on the Enslaved.org formal for the Enslaved.org Ontology were created using a sophisticated ontology and knowledge management system developed This provides a and for knowledge models a to use allowing researchers to define and relationships without the documentation and its we and This information was integrated using the specialized described in the A on Ontology with Ontology The and Ontology A The the process and for the of Ontology a form of standardized that documents the components and their These as an intellectual record, crucial and and of the knowledge the process and the Enslaved.org ontology, this section of the and modules of the Enslaved.org ontology. These are from the technical documentation for the Enslaved.org Ontology of information about an is as a record for that as in Each will have many records with it, and the collection of them everything that we know about that Each record by one of information about the information about this of has a of which are described the is that every record is of one of the by these is a of for records that to and not to in such as For a a of one of information, about the of the from one the of the record, provides the ability to about the document that an record is generated It also for that describe the of the data by different The of our in its formal structure precise relationships between historical documentation and knowledge. Each record functions as an with is and from one with This ensures of and in entities with may information about their sources, to one classification from a standardized This document the nature of the historical record, ship manifest, plantation ledger, baptismal which the information was initially The implementation of a vocabulary ensures across diverse and research These entities may practical or digital the archival repository where the document the database where its digital can be or to specific online This system helps with while bridges between archives and digital these entities may standardized information, also by a intellectual and This structured approach to management both and ethical of with historical particularly of this are from a data model of a diverse of contributors and their sources, while also multiple research and significant intellectual and technical address these and to data and for the we developed the modular Enslaved.org Ontology as our knowledge are significant for historical research on where ethical around data and access intersect with technical about how knowledge is and of in our modular ontology a fundamental truth about the historical record of every trace of an enslaved person's was as an baptism that an that human to a ship's that forced a that for an that the about These were not but when the of most the very documents that our evidence. While we can make our Enslaved.org Ontology work for the and with linking and machine we cannot allow the model to the of it is to record and search organizing our knowledge framework around events, we that historical documentation about enslaved people was and to specific when their lives with powerful to create This approach the that enslaved individuals between these when to capture their The modular researchers to connect these across time and building more of lives that records could only document but does not in way a of the whole individual key to it is for us to document our implementation of modular ontology methodology in the Enslaved.org a detailed that may similar in historical documentation and digital as as another of to the project. By our and we hope to to an of around structured knowledge in historical we have data expressed a modular the the of human to people historical systems to to economic When Maria baptismal record in Rio can connect to work history in Bahia and manumission in São we witness not just data integration but the of a human story that slavery to The ontology becomes both a technical and an of historical digital that communities to trace and family connections across while to questions that isolated databases could never work that methods can building not just better data systems but more of and The modular framework ensures this knowledge structure will as new the and new archives a for digital scholarship that both technical and the to history to For documentation, technical and contextual on the Enslaved.org are to our detailed resources