A framework for aggregation of presence information based on user provisioned rules

Presence systems are widely used to get aware of other users’ availability and willingness to communicate before actually contacting them. While early presence services offered only textbased communication and relied on the users to enter appropriate status descriptions, modern applications have evolved that integrate multi-media communication and facilitate automatic detection of status changes. To achieve this, unstructured data provided from various sources is aggregated and set into relation with user-specific context information. As most of the algorithms used for automatic status inference require extensive information acquisition to construct an initial rule set describing the user’s habits and activities, several research groups have investigated mechanisms for transforming low-level sensor data into abstract presence descriptions. In this thesis, we enhance the sensor-fusion approach to aggregate abstract presence descriptions from multiple presence-zones a single user might be part of. We propose a vocabulary based on ECMAScript for controlling the aggregation process. With this language, users can specify how status update notifications generated by their personal presence sources should be aggregated at a presence server, and how this information should be published to subscribed watchers. The latter allows for explicitly choosing specific communication channels offered to a particular set of users. For example, the mobile phone might be offered as a means for communicating with family members while this media is not offered to any other group of subscribers. To prove the applicability of our approach in specific environments, we have implemented a presence server that supports the evaluation of presence aggregation scripts as defined in this thesis. In addition, the standardized presence format was extended by a set of attributes that describe the degrading exactness of presence values over time. Based on this information, a user’s presence status can be calculated by a presence server even if no recent status updates have been received.