Computational ontologies (Pease 2011; Arp et al. 2015) are a key feature of data-sharing and data-labelling practices on the internet. Ontologies help integrate disparate or unorganized data to produce meaning, sort of “like a thesaurus, a finite set of terms, organized as a hierarchy that can be used to provide a value for an element” (Pomerantz 2015). Modern ontologies are an outgrowth of early artificial intelligence research in expert systems (Hayes-Roth et al. 1983) and knowledge representation (Sowa 1999). Today, many data-driven media technologies like virtual assistants and social media platforms use ontologies (Tecuci et al. 2016).
More specifically, media technologies like Google and Facebook’s graphs, semantic web standards like the World Wide Web Consortium’s Web Ontology Language (OWL), and virtual assistants like Siri, Cortana, Alexa, and Bixby provide unique opportunities for harnessing disparate data to increase knowledge mobilization via ontologies. However, like any technology, they can impede social progress if during their development designers are not also attentive to data justice issues. Ontologies present truly unique problems—they are not only a matter of quantification and sorting but also a matter of meaning. What counts as a restaurant in Siri’s Active Ontology? How are social entities and relations defined in OWL? What languages do ontologies recognize?
Building and extending ontology work to data justice and social progress issues involves looking at how ontology is connected to data gathering, data modeling, databases, metadata, and how the use of these and other tools like application programming interfaces (Helmond 2015) impact civil society through public facing ontology-driven apps and technologies. Drawing on the work of Gitelman (2008), Srinivasan (2012) offers one such approach by asking how we might include computational ontology in our discussion of “ethical questions about the sovereignty of diverse knowledge, and whether the voices of emerging users should be ignored or empowered” (205).
What happens, for example, when digital objects represent social entities and relations (Kallinikos et al. 2010; Hui 2012; Krämer and Conrad 2017)? This is a modern update to an old problem, one we have seen in critical scholarship on the history of the census, statistics, and, more recently, big data (Hacking 1982, 1991; Beer 2016). The question “who counts?” can be read as a double articulation—who is doing the counting and who deserves to be counted? Data ontologies are an update to those problems, complicated by semantics (“who counts what?”). Currently employed in areas as diverse as municipal administration, virtual assistants, scientific knowledge sharing, production and logistics, and intelligence gathering, data ontologies that deal with social entities and relations necessitate what Couldry and Kallinikos describe as a “new ontology of the social” (2017: 153). Computational ontologies encourage the datafication of social entities and relations by constructing social ontologies (Searle 2006) to provide labels for data in organized, semantic structures. Once completed, one may combine and analyze heterogeneous data in ways previously impossible when they retained their own idiosyncratic labels, and computations can extract new information.
To set the stage, the first paper in this panel describes the upper level Ontology Industry. Drawing on in-depth, long form, unstructured ethnographic interview data collected from multiple senior stakeholders in a variety of ontology projects, including developers in the private sector and researchers at nonprofit organizations, the paper describes several global ontology initiatives, users, and potential to impact civil society.
The second paper discusses several formal ontology development activities being carried out within the broader polar community. The project “Mapping the Arctic Data Ecosystem” aims to develop a formal ontology and network model of the Arctic data system. Technical relationships are documented as are data sharing and financial relationships. The paper provides a critical analysis of observed problems, risks, and benefits of the formal ontology projects described.
The third paper provides an analysis of city ontologies and homelessness. It presents the results of primary research conducted as part of a critical data and software studies project carried out in Dublin, Boston and Ottawa. The study examined how digital data were materially and discursively supported and processed in three homeless intake and case management systems, PASS, HIFIS and HUD HMIS compliant systems and how these systems ‘made up’ homeless people.
The fourth and final paper provides a broader media theory of the ontological challenges that arise when ‘the social’, or at least particular important sites for sociality and the production of social knowledge (including ‘social media’) are computed: that is, constituted by and through the outcomes of deep forms of data processing driven by instrumental practices of control and/or profit making.
