Mastering the Spatio-Temporal Knowledge Discovery Process

Contributions:

The PhD thesis addresses the followinf topics: a framework for supporting the progressive combination of mining, querying and reasoning on spatio-temporal data collected by  mobile devices, which represent trajectories of moving objects. The contributions of the thesis may be summarized as follows:

1. The definition of the theoretical framework for a novel data mining query language for trajectories, based on an elegant principle of compositionality of data and models which accounts for the combination of querying and mining.
2. The realization of a comprehensive software system that realizes the above framework based on an extensible modular architecture that supports the efficient execution of querying and mining operations as well as the extensibility of the systems with new querying and mining operators.
3. The invention and realization of two novel analytical methods: a preprocessing method aimed at trajectory reconstruction and a prediction method aimed at predicting the next location of a trajectory.
4. The invention and realization of a semantic component that allows the system to support the use of domain ontologies for automatic reasoning on trajectory data.
5. An experimentation on a real case of human mobility, consisting of  a massive data sat of GPS trajectories that has demonstrated both the analytical utility and its efficiency and scalability.

The system proposed in the thesis has been successfully presented in the final review of the European project GeoPKDD – Geographic Privacy-Aware Knowledge Discovery and Delivery, and has been presented as a representative of the EU FET program at the European Parliament in Strasbourg at an exhibit to the members of ITRE, the EU Parliament Committee on Industry, Telecom, Research and Energy (April 20-21, 2010).

Pdf download

On-line slides