Abstract
This paper proposes a method to represent the characteristics of a place (i.e., use of the venue, atmosphere of the area) by using geo-tagged microblog posts around the place. It enables a vector representation of a location similar to the distributed representation of a term in Word2Vec. Our method uses a simple neural network that is trained through the task of estimating the terms that appear in tweets posted from the area. The effectiveness of our method is illustrated through an experiment of a comparison of similar locations in Tokyo and Kyoto.
Author supplied keywords
Cite
CITATION STYLE
Shoji, Y., Takahashi, K., Dürst, M. J., Yamamoto, Y., & Ohshima, H. (2018). Location2Vec: Generating distributed representation of location by using geo-tagged microblog posts. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11186 LNCS, pp. 261–270). Springer Verlag. https://doi.org/10.1007/978-3-030-01159-8_25
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
