Semantic trajectory computation is the process of enriching raw mobility data with contextual information in order to extract knowledge for decision maker. In this paper, we present the different works tackling the semantic amelioration of mobility data description. We focus on introducing a taxonomy according to the techniques used and the types of semantics added. We classify literature into three main categories. The raw mobility data based trajectory annotation, where machine learning and statistics are used to infer moving entities’ activities from raw mobility data without any additional semantic information. The static trajectory annotation where geographic information, stored in spatial databases, is explicitly associated with raw mobility data. The dynamic trajectory annotation where Web resources use has become widespread across the globe, making it an unavoidable source of spatiotemporal information that describes the movement and activities of human beings. Once raw mobility data enriched with semantic information, knowledge discovery techniques are employed to extract knowledge.