The next speaker in this IAMCR 2019 session is Andrew Iliadis, whose focus is on the role of metadata. Metadata and related terms such as ontology have rocketed to broader attention in recent years; here, philosophical concepts related to ontology have come to be translated to computationally accessible relationship constructs between data entities.
This renewed interest in ontologies is related to the growth in available data from a wide variety of sources; we are now at an advanced point in the hype cycle for computational analysis for such data, and further advances require better tools for connecting these disparate and disconnected datasets.
But this implies particular logics in our understanding of data and metadata, and the choices made here may also affect platform affordances and user practices and perspectives towards data. For instance, the W3C works with a subject-predicate-object philosophy in the development of its Semantic Web ontologies; other approaches implement different forms of knowledge graphs to represent their data.
Virtual assistants and recommendations engines, in particular, rely on applied metadata ontologies to sort the datasets available to them and make them interoperable. Some of these are sketched out in the public engineering blogs related to these products, and a critical aspect of such systems is also the contextualisation of the available knowledge graphs to the user engaging with them.
For some of these projects, one aim is also to replace actual Websites with the information stored in the knowledge graph itself – as when Google provides the answers to specific questions right in the search engine, rather than simply pointing to links with more information. This is especially the case with voice-driven virtual assistants, of course.
The next step in this process, then are industry-wide knowledge graphs that even span the data held by companies that are otherwise in competition with each other. This is the latest step in a historical trajectory from micro-level ontology frameworks like XML through meso-level frameworks like Dublin Core to the latest macro-level frameworks. It will be critical to examine the politics built into such ontology frameworks.
