Extracting localized information from a Twitter corpus for flood prevention

TL;DR

This paper presents a method for analyzing Twitter data related to tropical storm Harvey, focusing on geographic accuracy and topical content to aid flood prevention efforts.

Contribution

It introduces a framework for collecting, spatially analyzing, and processing unlabeled Twitter data for disaster-related information extraction.

Findings

Assessment of geographic information quality in Twitter corpus

Strategies for processing unlabeled tweets

Insights into topical content representation

Abstract

In this paper, we discuss the collection of a corpus associated to tropical storm Harvey, as well as its analysis from both spatial and topical perspectives. From the spatial perspective, our goal here is to get a first estimation of the quality and precision of the geographical information featured in the collected corpus. From a topical perspective, we discuss the representation of Twitter posts, and strategies to process an initially unlabeled corpus of tweets.