Over the last years, computer networks have evolved into
highly dynamic and interconnected environments, involving multiple heterogeneous
devices and providing a myriad of services on top of them.
This complex landscape has made it extremely difficult for security administrators
to keep accurate and be effective in protecting their systems
against cyber threats. In this paper, we describe our vision and scientific
posture on how artificial intelligence techniques and a smart use of security
knowledge may assist system administrators in better defending
their networks. To that end, we put forward a research roadmap involving
three complimentary axes, namely, (I) the use of FCA-based mechanisms
for managing configuration vulnerabilities, (II) the exploitation of
knowledge representation techniques for automated security reasoning,
and (III) the design of a cyber threat intelligence mechanism as a CKDD
process. Then, we describe a machine-assisted process for cyber threat
analysis which provides a holistic perspective of how these three research
axes are integrated together.