In the forefront of efforts to curb energy consumption and as a consequence decrease greenhouse emissions, cities as well as individuals, turn to the field of Smart Homes to optimize their heating schedules through IoT-enabled solutions. However in many cases efforts are focusing on algorithms and systems requiring large amounts of processing power and constant data availability to be effective. In this paper, an approach tailored to the constrained resources in the IoT domain is introduced that is based on the Social IoT paradigm, instead of centralized computational nodes. The framework enables Smart Home Gateways to seek solutions to their heating schedule needs through communication of actual observations with fellow homes rather than brute force calculations based on probabilistic models that may require centralized approaches. Using the provided IoT components of the COSMOS ecosystem, Smart Homes may run purpose-built applications that use stored Knowledge, communicate it throughout the Network of Things and act on it in ways which aid the end users in retrieving relevant solutions. Raspberry-based simulations indicate that this diffusion of Knowledge as well as the improvements and evaluations through feedback performed on it, allow for the creation of a lightweight and resource effective approach, on the problem of Heating Management.