http://lattes.cnpq.br/5704323329174608; ALMEIDA, Carlos Augusto de Santana.
Resumo:
The popularization of location-aware mobile devices such as mobile phones and GPSs has enabled large-scale monitoring of mobile objects carrying such devices, such as people, cars, and airplanes. This monitoring results in the generation of a large mass of raw data about the trajectories of these objects. The analysis of this type of data allows us to discover behavioral patterns that can be explored in a large number of applications, such as urban traffic management, the study of tourist trajectories in a trip, to identify bird migratory routes, among other applications.
To turn this raw data mass into useful information for decision making and knowledge discovery, a suitable way would be to make it available in a Data Warehouse (DW), a database optimized to handle large volumes of data efficiently. For conventional data, DW technologies have been used successfully for decades. However, the nature of trajectory data poses certain challenges for building and maintaining DW, including: (i) the large amount of trajectory data consumes a lot of memory and processing resources, making query execution time too long, making it impossible OLAP style analysis; (ii) support provided by DW technologies for trajectories is still limited to the storage and retrieval of trajectory data, OLAP operations cannot be performed on them as is the case with spatial data; (iii) scarcity of models, there are few studies related to multidimensional modeling for trajectories. In order to solve part of these problems, this work proposes: (i) a multidimensional spatial model capable of providing OLAP analysis for trajectories (roll-up for trajectories), thus allowing to analyze the behavior of moving objects over and between regions in space and time; (ii) furthermore, the model allows segmenting trajectories into various semantic components, which can carry information that gives meaning to trajectory; and (iii) provide trajectory compression, a way to reduce the amount of data stored.