Knowledge Representation in Practice : Advanced knowledge-base interactions in industrial applications

There is a growing insight that a key factor for the success of an AI system is often its ability to interact in a meaningful way with users. In Decision Support applications, communication with the decision makers is key: after all, if they do not understand the system, they will be reluctant to follow its advice. In this PhD, we will improve on the current state-of-the-art on this topic in two ways. First, there is a recent evolution towards allowing knowledge bases to be written and maintained by domain experts directly, cutting out the knowledge engineer as a middle-man. A prime example of this is the recent Decision Model and Notation standard, which is rapidly gaining traction in industry. We will examine how to extend the expressivity of this standard, such that it becomes able to handle all the knowledge needed for automated engineering applications. Second, users must also be able to interact with an existing knowledge base in intuitive ways. Where traditional systems typically offered only the functionality, e.g., computing a complete design, modern systems also offer functionality such as explaining why specific part of a design is (not) possible, identifying which design choices are relevant in a certain context, etc. The IDP knowledge-base system of DTAI group plays a leading role in research on such interaction methods. We will extend the currently available methods which new logical interactions that are aimed at helping the designer understand and structure the (possible huge) design space, given the available knowledge. For both tasks, we will investigate the possibility of enriching the expert knowledge with the results from a data-based analysis.