Context-aware systems are an instance of the ubiquitous or pervasive computing vision. They sense the users' physical and virtual surrounding to identify the best system support for that user and adapt the system behaviour accordingly. The overall architecture of a context aware system can be broken into a number of logical aspects: gathering context data, storing the data, deriving knowledge through reasoning and mining and retrieving that knowledge to finally adapting system behaviours. Context is anything characterizing the environment of the user -their location, the ambient temperature, the people they are with, their current activity and some even consider the user's mood. Traditionally context information has been gathered through the use of hardware sensors, such as GPS sensors or smart badges for locations and there has been work to track user's eye movements at their desk to see which application they are using. However, determining the activity of a user has shown to be elusive to being sensed with hardware sensors. As users use web services more frequently they are exchanging messages with the services through the SOAP protocol. SOAP messages contain data, which is valuable if gathered and interpreted right -especially as this data can be shedding information on the activity of a user that goes beyond "sitting at the computer and typing". We propose a complimentary sensor technology through the use of software sensors. The software sensors are essentially based on monitoring SOAP messages and inserting data for further reasoning and querying into a semantic context model. In this paper we consider details of extracting the data from SOAP messages in a non-obstructive way and show a solution to map the data from a SOAP message to our OWL ontology model automatically. On the latter, we specifically explain the methodology to map from SOAP messages to an existing structure of knowledge.