This session included four fascinating presentations, focussing on different aspects of Knowledge Representation and Reasoning. The first paper, titled ‘Constructing Knowledge Graphs of Depression’, was presented by Frank van Harmelen, and was co-authored by Zhisheng Huang, Jie Yang, and Qing Hu. Frank presented a collaborative project between the VU and Beijing Anding hospital that focused on constructing a large knowledge graph of depression by integrating various knowledge resources using semantic web technology. Due to its high degree of inter-operability, this approach enables psychiatric doctors to efficiently find answers to queries without having to explore multiple databases.

The second paper, titled ‘Using Values and Norms to Model Realistic Social Agents’, was presented by Rijk Mercuur, and described joint work with his supervisors at Delft University of Technology, Virginia Dignum and Catholijn Jonker. The presentation addressed an interesting study that aimed to evaluate the benefit of values and norms for developing more ‘socially realistic’ agents. By simulating a psychological experiment known as the Ultimatum Game with three different types of social agents, the authors found that agents using a theory that combines both values and norms produce behaviour that is most similar to the behaviour of humans.

The third paper, titled ‘Evaluating Intelligent Knowledge Systems’, was presented by Neil Yorke-Smith of TU Delft. In his engaging talk, Neil shared his experiences from a large seminal project on the development of a user-adaptive personal assistant agent for time management assistance. The presentation focused on the (often underestimated) role of evaluation in such a project, and concluded with several useful ‘lessons learned’, of which the main message was that researchers and project managers benefit from adoption of best practice in evaluation methodologies from the start of a technology project.

The fourth paper was titled ‘On the Problem of Making Autonomous Vehicles Conform to Traffic Law’ and was presented by Henry Prakken (Utrecht University & University of Groningen). As the title suggests, the presentation addressed the problem of how we can make the behaviour of self-driving cars conform to traffic laws, although it could also be seen as a special case of the general problem of making intelligent autonomous systems conform to the relevant laws. Henry discussed various features that make traffic law challenging for AI & Law models (such as exceptions, rule conflicts, and the need for common-sense knowledge) and evaluated three approaches to the design of law-conforming autonomous vehicles in light of these challenges. All in all, this was a very lively session with a lot of interesting discussions.