Geo-spatial Trend Detection through Twitter Data Feed Mining

Abstract

Present-day Social Networking Sites are steadily progressing towards becoming representative data providers. This paper proposes TweetPos, a versatile web-based tool that facilitates the analytical study of geographic tendencies in crowd-sourced Twitter data feeds. To accommodate the cognitive strengths of the human mind, TweetPos predominantly resorts to graphical data structures such as intensity maps and diagrams to visualize (geo-spatial) tweet metadata. The web service’s asset set encompasses a hybrid tweet compilation engine that allows for the investigation of both historic and real-time tweet posting attitudes, temporal trend highlighting via an integrated animation system, and a layered visualization scheme to support tweet topic differentiation. TweetPos’ data mining features and the (geo-spatial) intelligence they can amount to are comprehensively demonstrated via the discussion of two representative use cases. Courtesy of its generic design, the TweetPos service might prove valuable to an interdisciplinary customer audience including social scientists and market analysts.