Smart environments, enabled by the Internet of Thing (IoT) paradigm, advocate for more intelligent and interconnected systems, electronic devices, tools, and appliances. While most efforts are nowadays addressed to provide connectivity or smartness to IoT devices, unfortunately, few have realised the importance of supporting automatic service composition and service reconfiguration capabilities at middleware level. Due to the openness that characterise such environments, the range of possible interactions and available services and devices cannot be totally defined nor prescribed in advanced. This uncertainty demands mechanisms to dynamically adapt existing systems, and their functionality, to address unforeseen needs. In order to do so, a general understanding of contexts, services, and device capabilities is needed. From that understanding, new responses can be automatically devised on run-time. This paper presents a semantic middleware specifically devised to support automatic and autonomous service reconfiguration and composition. The novelty consists in moving the semantics, traditionally held at the programming-interface level, to a common-sense knowledge-base system, with higher expressive power and reasoning capabilities.