The seminar brought together researchers and domain experts involved in developing and utilizing methods for knowledge extraction from massive amounts of data from moving objects. This knowledge is essential to substantiate decision making in public and private sectors, in application domains such as transportation modelling, urban planning, tourism, wildlife ecology, spatial epidemiology, location-based services, flight safety, and marine safety. Moving object data typically include trajectories of discrete spatial objects (e.g. humans, vehicles, animals, and goods), continuous phenomena (e.g. storms, ocean currents) as well as trajectories of abstract concepts (e.g. information flow, moving data points in attribute space) or even vectors of spreading diseases. Technologies for object tracking are low cost and increasingly reliable in terms of coverage and accuracy, hence movement records are nowadays generated in huge volumes on a routine basis, using diverse technologies such as radio frequency mapping, Global Navigation Satellite Systems, video sequences and Doppler radar.

The computational analysis of movement data has seen a successful first decade with progress made in capturing, preprocessing, storing, indexing and querying movement data, combined with promising results in visualizing movement and detecting movement patterns. However, whereas such basic progress in handling movement data was needed for establishing a new field and attracting funding, attention must now move on towards the extraction of useful information and process knowledge from tracking data.

In many application fields the need for analysing large sets of trajectories is evident and crucial; however, only very rudimentary automated analysis tools are available and anything more advanced is currently analyzed manually. As an example there are several companies tracking the movement of football players through video with a frequency of at least 25Hz and accuracy of approximately 10cm. But most of the analysis and the annotation (passes, throw-ins, goals etc) of the data are still made manually. Thus the analysis part has been neglected and in comparison with the image processing part it is technologically far behind in the development. A reason for this is the obvious lack of theoretical and practical solutions for many crucial fundamental problems.

For that reason, this seminar focussed on formalizing methods for algorithmic analysis, visual analytics, data mining and knowledge discovery, defined by a multidisciplinary team of researchers and practitioners.

Participants and Format

The seminar brought together researchers from several disciplines involved in developing and utilizing computational techniques for spatiotemporal object representation, data mining, and visualization. This community encompasses an interdisciplinary mix of methodologically oriented as well as application oriented researchers. The methods-oriented researchers are from fields such as theoretical computer science, spatial databases, knowledge discovery and data mining, visual analytics, and geographic information science. They were complemented by application scientists from various fields, especially behavioral ecology and urban planning.

Drawing upon positive experiences in previous seminars of this series, oral presentation sessions were complemented by special sessions dedicated to open research questions and project ideas, as well as to discussions in small, concurrent break-out groups focussing on a specific domain or case studies. Since the participants of the seminar came from quite different backgrounds, concise survey talks on the first two days were delivered on movement ecology (Ran Nathan), industry movement research (Ben Loke) and spatial data mining (Gyözö Gidófalvi).

A data challenge was organized for the participants prior to the seminar. This ensured that the participants were well aware of the application domain which was the focus of the seminar and it gave the domain experts a possibility to see the potential use of various different approaches. The data challenge included a bird migration data set provided by Emiel van Loon.

Many interesting research results were presented, demonstrating the progress in this field. The participants were highly satisfied with the quality of the seminar; especially the involvement of domain specialists from relevant application domains was highly appreciated.