On-line Geospatial Term Extraction from Streaming Geotagged Tweets

---

Tweet data

Date: 30days(2016/09/08~2016/10/09)
Tweet: 6,655,763 geotagged tweets
*We removed tweets of two datas(10/03, 10/05) because we was not able to get tweets.

geospatial terms

stationary term
temporary term
Altough our proposed method analyse tweet data on real-time, we demonstrate geospatial terms which are included in the each database at the end of each day here.
The location of the geospatial term was estimated from its set of the geotags by using Kernel Density Estimation.

GeoNames

http://www.geonames.org/
GeoNames contains the names of over 2.2 million places in the United States with their lat/long coordinates. These entries have been provided by official public sources, and additionally, have been added/edited manually by users. Each place can be associated with more than one lat/long coordinates.

Cheng's mthod

This method constructed static GTDb applying batch method to a set of geotagged tweets posted during the 1st 10 days. The trained classifier selected 10,082 geospatial terms. Additionally, 7,322 city names in the Gazetteer are manually added to construct a GTDb with 17,404 geospatial terms in total.
Z. Cheng, J. Caverlee, and K. Lee, "You are Where You Tweet: A Content-based Approach to Geo-locating Twitter Users," Proc. of ACM Int'l Conf. on Information and Knowledge Management, pp.759-768, 2010.