About This Project

Location enabled mobile devices are increasingly used to share data within the online community. Web and mobile applications facilitate the crowd-sourcing of geo-data, such as tweets, photos and map data. We analyze a great variety of data sources mined from the internet to see how people contribute to these services. The goal is to gain a better understanding of human movement, data quality and the usability of crowd-sourced data in real word applications.

Ask the Scientists

What is the context of this research?

GPS enabled smart phones used on Web 2.0 applications and social media platforms generate an abundance of crowd-sourced geodata, such as tweets, flood images, or digitized roads, contributed by many civil-minded enthusiasts. Quality assessment of such crowd-sourced datasets, often referred to as Volunteered Geographic Information (VGI), requires an understanding of how data from different platforms are cross-linked, and how user communities evolve across platforms. We analyzed in a conference paper the geographic overlap of data contributed to OpenStreetMap (a digital fine scale world map) and Mapillary (a collection of street level photographs). We pushed this research direction as co-organizers of a recent international workshop.

What is the significance of this project?

The participation in big data by people globally provides massive geo-datasets which can be processed, quality-controlled, and ultimately applied in facilitating emergency response, transportation modeling, participatory planning, or citizen science. Since anyone can add anything to this big data pool without any control, a major challenge to extract meaningful information is to understand how people would use these services. Analyzing contributions to several VGI and social media platforms, our project will recommend scientific methods for a better filtering of quality data from noise, and therefore, guide the proper use and analysis of crowd-sourced geodata.

What are the goals of the project?

Using an exploratory approach, the project aims to discover the advantages of analyzing patterns in multiple platforms in order to understand user behavior, as opposed to analyzing single platforms. The main goals are to determine how users participate in several services, and to provide recommendations of data use for applications that use VGI. Using already downloaded data (e.g. 4 billion geotagged tweets) and user data through Web services, we analyze spatial overlap, geographic clustering, and data volume to understand how similarly a user uses the space in different online platforms. We test if cross-referenced data have better data quality than others. We share tools with the public on GitHub, extending our previous work.