Context-aware applications play a central role in Ambient Assisted Living (AAL) environments. A study about ethical, legal and social implications of AAL environments revealed that AAL applications need to fulfill two central requirements:(a) they must by tailored towards the specific needs of impaired persons, and (b) they need to be easily configurable by different user groups such as caregivers and relatives since these groups usually do not have profound technical background knowledge. This work discusses how the two main results of a study about ethical, legal and social implications of AAL environments, namely (a) the need-centric orientation of AAL applications and (b) their configurability by non-technical persons, can be addressed by a technical framework that uses semantic Web techologies as interoperabiltiy infrastructure for the integration of IoT devices. The architecture addresses typical aspects of AAL. We show how aspects related to the extensibility of AAL environments and error-checking methods can be applied by the presented approach. More specifically, we present a semantic model to describe user intentions about IoT devices as well as a method to reason about user intentions and to semi-automatically evoke appropriate actions. Our approach is based on the idea of the Web of Things (WoT) and evaluated using a scenario about helping an impaired person with controlling devices.