Computational inference of conceptual trajectory model : considering domain temporal and spatial dimensions. (Raisonnement sur la modélisation des trajectoires : prise en compte des aspects thématiques, temporels et spatiaux)
暂无分享,去创建一个
Spatio-temporal data describing trajectories of moving objects has increased as a consequence of the larger availability of such data due to current sensors techniques. These devices use different technologies like global navigation satellite system (GNSS), wireless communication, radio-frequency identification (RFID), and sensors techniques. Although capturing technologies differ, the captured data has common spatial and temporal features. Thus, relational database management systems (RDBMS) can be used to store and query the captured data. RDBMS define spatial data types and spatial operations. Recent applications show that the solutions based on traditional data models are not sufficient to consider complex use cases that require advanced data models. A complex use case refers not only to data, but also to the domain expert knowledge and others. An inference mechanism enriches semantic trajectories with this knowledge. Temporal and spatial reasoning are fundamental for the inference mechanism on semantic trajectories. Several research fields are currently focusing on semantic trajectories to discover more information about mobile object behavior. In this thesis, we propose a modeling approach based on ontologies. We introduce a high-level trajectory ontology. The temporal and spatial parts form an implicit background of the trajectory model. So, we choose temporal and spatial models to be integrated with our trajectory model. We apply our modeling approach to a particular domain application : marine mammal trajectories. Therefore, we model this application and integrate it with our ontology. We implement our approach using RDF. Technically, we use Oracle Semantic Data Technologies. To accomplish reasoning over trajectories, we consider mobile objects, temporal and spatial knowledge in our ontology. Our approach demonstrates how temporal and spatial relationships that are common in natural language expressions (i.e., relations between time intervals like ”before”, ”after”, etc.) are represented in the ontology as user-defined rules. To annotate data with this kind of rules, we need an inference mechanism over trajectory ontology. Experiments over our model using the temporal and spatial reasoning address an inference computation complexity. This complexity is indicated in term of time computations and space storage. In order to reduce the inference complexity, we propose optimizations, such as domain constraints, temporal and spatial neighbor refinements. Moreover, controlling the repetition of the inference computation is also proposed. Even more, we define a refinement specifically for the application domain. Finally, we evaluate our contribution. Results show their positive impact on reducing the complexity of the inference mechanism. These refinements reduce half of the time computation and allow considering bigger size of the data.