Addressing the Sparsity of Location Information on Twitter

Micro-blogging services such as Twitter have gained enormous popularity over the last few years leading to massive volumes of user generated content. In combination with the proliferation of smart-phones, this information is generated live from a multitude of content contributors. Interestingly, the content and timestamp of tweets is not the only information that can produce useful knowledge. The location information of users is of great significance since it can be utilized in a variety of applications such as emergency identification, tracking the spread of a disease and advertising. Unfortunately, information regarding location is very rare since many users do not accurately specify their location, and fewer posts have geographic coordinates. In this work, we aim to confront this data sparsity issue. Utilizing Twitter’s social graph and content, we are able to obtain users from a specific location. We optimize our method to work with minimum amount of queries considering the large volume of data in such settings. We also provide a mechanism for geo-locating a tweet within a city and present the qualitative enrichment in our data, achieved by our method.