Uncertainty- and Trust-Aware Integration of VGI andSpatio-Temporal Traces for Understanding Animal Behavior

点击翻译按钮获取中文摘要

The popularity of location-sensing devices (e.g., sensors and smartphones) and online platforms (e.g., eBird.org and ornitho.de) facilitates the generation of large spatial data sources also known as Volunteered Geographic Information (VGI). The heterogeneous data provided voluntarily from citizens contains, for instance, GPS data, images, annotations, etc. The large VGI sources contain valuable information, but the processing and analysis of VGI are challenging due to their data quality, varying level of detail, uncertainty, and heterogeneity. There is a lack of methods to combine VGI with other data sources and a lack of scalable interactive geo-visualizations to investigate them. In this proposal, we therefore plan to develop novel Visual Analytics methods to semi-automatically integrate and analyze large VGI datasets with spatiotemporal tracking data. We will develop new methods to combine the huge amounts of data from birdwatching platforms (e.g., ebird.org and ornithon.de) and the ICARUS (International Cooperation for Animal Research Using Space) project to investigate behavioral and movement traits of animals. There is an immense value in combining the human-generated VGI data providing semantic context and understanding with the accuracy and reliability of the tracking data provided by the ICARUS system. The combination will help to tackle numerous important questions that are difficult to answer without integrating the VGI and Non-VGI information. In particular, our system will help to investigate local animal habitats, biodiversity loss, animal migrations across continents, land-use change, invasive species, the spread of diseases, and ultimately climate change.Currently, there is no system that can automatically integrate and analyze these massive heterogeneous data sources while tackling the necessary uncertainty and trust aspects. We propose to follow the Visual Analytics approach to combine the strengths of humans and computers to generate knowledge from these large data sources. The developed Visual Analytics system will help to interactively integrate, match, annotate, and analyze models generated from the VGI and animal tracking data.