Georeferencing Flickr resources based on textual meta-data

摘要: The task of automatically estimating the location web resources is central importance in location-based services on Web. Much attention has been focused Flickr photos and videos, for which it was found that language modeling approaches are particularly suitable. In particular, state-of-the art systems georeferencing tend to cluster locations Earth a relatively small set disjoint regions, apply feature selection identify location-relevant tags, then use form text classification area most likely contain true resource, finally attempt find an appropriate within identified area. this paper, we present systematic discussion each aforementioned components, based lessons have learned from participating 2010 2011 editions MediaEval's Placing Task. Extensive experimental results allow us analyze why certain methods work well show median error just over 1km can be achieved standard benchmark test set.