Social media has become a useful data source for processes and researches interested in improving people's life. Publications done by social media users provide details about the people perceptions of their environment, and updated observations about what happens in real world. In this approach the relevance of Twitter as a data source for scientific purposes is analyzed, as well as its use in geospatial researches. Tweets have two main characteristics, a text where user describes its ideas, and metadata, where features such as the coordinates of the place where the tweet was posted are stored. Different computing procedures are applied over tweets in order to make them useful for different tasks; commonly, text mining, classification, and regression algorithms are used to process tweets. The coordinates of tweets make possible to link events described in tweets with the geographical area where they occur. The analysis of tweets and coordinates provides updated data, useful in natural disasters control, decision taking processes and urban studies. The present approach studies what motivates users to tweet, and analyzing the messages produced by Twitter users, classifies the tweets into three groups according to the information expressed in their text: self-centered, social information and collective information. Additionally, some methods to extract information from tweets are studied, common problems presented when working with tweets, and some researches that use them in a geospatial domain are presented.

No abstract