ABSTRACT

The early inference controllers were developed in the 1990s (Thuraisingham et al. 1993). These inference controllers processed security policies, which were called security constraints at that time, during query, update, and database design operations. However, the early work in the area of MLS reasoning was rather limited due to inadequacies of policy representation and reasoning techniques. Ideally, an inference controller should be able to detect inference strategies that users utilize to draw unauthorized inferences and consequently protect the knowledge base from such

security violations. Different approaches can be employed for building an inference controller. For example, it is desirable to use state-of-the-art machine-learning techniques to build a learner that automatically learns to recognize complex patterns and make intelligent decisions based on some input. One could also build an inference controller that uses Semantic Web technologies that are equipped with reasoners that perform inferences over the facts in a knowledge base. Semantic Web technologies have overcome the significant limitations that were present in the 1990s. In particular, our research has focused on designing and developing an inference controller with sophisticated reasoning and policy representation techniques using Semantic Web technologies. This chapter will describe the detailed design and implementation of an inference controller that draws inferences from a provenance graph store.