The purpose of this demo is to show our solution on how to find relevant answers to mistyped spatial-keyword queries. It is built using a multimedia metadata collection extracted from Flickr pages, called CoPhIR Test Collection
We processed the dataset to extract the photos taken in the U.S. based on their latitude and longitude values. Moreover, we used the keywords in the title, description, and tags of a photo as its textual attribute. The final dataset had about two million objects. Each record had a URL corresponding to the photo or a page including the photo. The total data size was about 400MB.
Fuzzy Keyword Search on Spatial Data Using the GeoNames Dataset