Anomaly detection is a key feature of applications processing singularities using IoT sensor measures. To guarantee high quality detections, meta-data providing spatio-temporal contexts on sensor measures are needed. In this paper, we introduce Scouter, a generic tool that helps in capturing, analyzing, scoring and storing the contextual information of a given application domain. The process depends on a semantic-based approach that exploits ontologies to score the relevancy of contextual information. The paper provides details on the system's architecture, describes its components and evaluates the performance based on real-world datasets.