Sensor networks provide means to link people with real world by processing data in real time collected from real-world and routing the query results to the right people. Application examples include continuous monitoring of environment, building infrastructures and human health.Many researchers view the sensor networks as databases, and the monitoring tasks are performed as subscriptions, queries, and alert. However, this point is not precise. First, databases can only deal with well-formed data types, with well-defined schema for their interpretation, while the raw data collected by the sensor networks, in most cases, do not fit to this requirement. Second, sensor networks have to deal with very dynamic targets, environment and resources, while databases are more static.In order to fill this gap between sensor networks and databases, we propose a novel approach, referred to as "spatiotemporal data stream segmentation", or "stream segmentation" for short, to address the dynamic nature and deal with "raw" data of sensor networks. Stream segmentation is defined using Bayesian Networks in the context of sensor networks, and two application examples are given to demonstrate the usefulness of the approach.