Teaching parliamentary procedure to machines: evaluating approaches to modelling in a complex domain.

UK Parliament is a highly complex domain. Designing knowledge graphs in complex domains presents a number of challenges. Many of the constraints do not arise from technology or design choices but rather from the complexity inherent in a loosely coupled democratic system which is necessarily flexible, adaptive and evolutionary. Our talk will emphasise the importance of non-technical approaches to succeed in such an environment.

We will outline the approach we took to creating ontologies for UK Parliamentary business and procedure, using bounded contexts to represent multiple viewpoints across a single problem space.

We will cover the use of domain-driven design, working in the open and Cynefin to mitigate the challenges presented by complex domains. We will also cover how we facilitate modeling sessions, some of the tooling we use, aspects of data governance and our approach to documentation.