# Text-based Geotagging — Data

来源：https://www.kaggle.com/competitions/geotagging-100/data

采集时间：2026-10-24T16:13:45Z

## The dataset provided is an:
- annotated corpus of 500k elements
- texts, as well as the respective geo coordinates (lat, long values)
- 123 regions covered
- 5000 tweets per location

The training data set provided in an archive containing 123 .json files, broken down by regions. For example, a single file could contain all texts from Brooklyn, while the other file would have all the texts posted from Tokyo & its suburbs.