Recent advances in data analytics prompt dynamic datadriven vulnerability assessments whereby data contained from vulnerabilityalert repositories as well as from Cyber-physical System (CPS) layer networks and standardised enumerations. Yet, current vulnerability assessment processes are mostly conducted manually. However, the huge volume of scanned data requires substantial information processing and analytical reasoning, which could not be satisfied considering the imprecision of manual vulnerability analysis. In this paper, we propose to employ a cross-linked and correlated database to collect, extract, filter and visualise vulnerability data across multiple existing repositories, whereby CPS vulnerability information is inferred. Based on our locally-updated database, we provide an in-depth case study on gathered CPS vulnerability data, to explore the trends of CPS vulnerability. In doing so, we aim to support a higher level of automation in vulnerability awareness and back risk-analysis exercises in critical infrastructures (CIs) protection.