The effects of buying a new car: an extension of the IDP Knowledge Base System

A long term goal in knowledge representation is the development of a knowledge based system (KBS). A KBS stores knowledge in a completely declarative way and is equipped with various domain independent inference methods to perform different reasoning tasks for a broad field of applications. In this paper we took a state-of-the-art KBS, IDP, with corresponding language FO(), and tested its applicability on a prototypical example from the business rule domain: a car rental system. We investigated whether we can indeed represent all required knowledge in FO(). The results are mixed: on the one hand, the expressible knowledge can be used to a greater extent (for example for profit maximisation), but on the other hand, some knowledge was not expressible in the current version of our system. One obstacle we encountered is that FO() does not have a good formalism to model the knowledge about situations where a new object is created. To remedy this, we propose an extension of the concept of inductive definitions which allows us to model this knowledge in a declarative way. As a result, this extension can be used for the car rental problem domain, as well as for many other applications and inference tasks.