Exploiting semantic extraction for spacial-temporal indexing in GEONODE

This paper presents an approach to information access based on spatiotemporal indexing. We describe GeoNODE, a “news on demand” system that visualizes news stories and other articles in terms of chronological time lines, geographical maps, and spatiotemporal animations. This spatiotemporal indexing is critically informed by our information extraction system, Alembic. This paper thus covers several dimensions of the technological partnership between GeoNODE and Alembic. We start with the end product, the display of spatiotemporally indexed documents, and work our way back through the various aspects of information extraction that underlie the indexing. We focus especially on those extraction tasks that operate at a common semantic level across diverse document sources and diverse languages.