Collaborative Georeferencing

---

    The goal of this project is to provide a mechanism whereby groups of users can form communities to collaboratively georeference and verify a shared dataset. This collaborative georeferencing framework consists of two end-user components:

1. The GEOLocate web-based collaborative client for reviewing and editing community records.
2. Web-based data management portal for creating and managing communities, their respective users and data sources

   Shared community datasets created via the portal may consist of multiple underlying data sources from either live DiGIR providers and/or uploaded text files. Support for TAPIR providers is currently under development. Data are stored using the full Darwin Core 1.2 specification, but subsets and/or alternatives schemas may be imported using the schema mapping interface. During import, data items are automatically normalized, georeferenced and related to one another via a similarity index. This index is used to identify all records that appear to describe the same collection locality regardless of syntax. During coordinate verification, users have the option to re-classify records that were incorrectly related to one another.

   Verification and correction of the computer generated geographic coordinates is accomplished using the GEOLocate desktop application. GEOLocate allows users to login to their communities, retrieve and visualize results, make any necessary corrections, provide additional comments, define errors as polygons, and save the results back to the shared dataset. The verified results of georeferencing can then be downloaded via the portal’s data management interface for re-import to the parent database.

   To examine the gains in efficiency over traditional georeferencing 2100 randomly selected collecting events from the TUMNH fish collection were imported and georeferenced using the collaborative georeferencing framework. The TUMNH fish collection was georeferenced by hand in the mid to late 90’s and therefore provides a useful test bed for assessing the efficiency and accuracy of automated methodologies. Of the 2100 records, 30% were identified as being similar to other records and an additional 33% were duplicates leaving a total of 782 unique locations requiring correction, a 63% reduction in effort overall.

Video Tutorials

• Using the Collaborative Georeferencing Web Client:
• Watch Online
• Download/Stream Windows Media Video Format (WMV) (appox. 50 MB)
• Download MPEG-4 Format (MP4) (appox. 50 MB)