This section presents some relevant projects for geographic information retrieval. The list is by not meant to be exhaustive, but gives some indications to the activity in the field in terms of projects working with georeferenced IR.
The first system to automatically index documents based on places name and phrases was GIPSY: The Goereferenced Information Procession System. [9] Gipsy used a three-step algorithm which relied on a gazetteer or thesaurus containing place names and the names of other geographically significant object. First all content-bearing geographic words and phrases were identified. Then these words and phrases were analysed and geographic coordinate data were assigned to them. Finally the resulting polygons from step one were stacked on top of each other, forming a topology. The peaks or areas above a threshold height of this topology were then chosen as the footprint for the document.
Other projects that have been oriented towards parsing text to find geographic references are the ADL Geospatial Integration Project [2] and the Berkeley Going Places in the Catalog Project [4].