The emergence of the Internet of Things has fueled a proliferation of smart things in many fields, including cultural spaces. Context-awareness addresses the production of large volumes of context by analyzing raw data and adding a meaning to them. Middleware systems have emerged, which perform context modelling and reasoning, supporting context-aware applications. The services provided by such applications can be personalized, automated and adapted to the current situation, thus enhancing the user interaction with the devices and the digital environment. In this work, a context-aware middleware system is presented, based on a hybrid reasoning schema, which combines multiple techniques to efficiently address each problem. The proposed middleware system is evaluated in a cultural space, where scenarios were designed and tested, using a mixture of real and artificial data. The experiments measured the accuracy, performance in terms of reaction time and scalability and the interactivity enhancement, achieved by the proposed middleware.