Tools for Mapping Twitter Conversations  

Vosoughi, S., Vijayaraghavan, P., Yuan, A., & Roy, D. (2017, May). Mapping Twitter conversation landscapes. Paper presented at the 11th International AAAI Conference on Web and Social Media, Montreal, Canada. Retrieved from http://hdl.handle.net/1721.1 /109773

In this paper, the authors detail the development and testing of an interactive, web-based tool that can be used to map the conversation landscapes on Twitter automatically. The system combines natural language processing advancements and deep, robust neural networks in conjunction with a scalable clustering algorithm and interactive visualization engine. This allows users to analyze the big data produced by Twitter to gauge public responses to issues and events. The system follows a four-stage “pipeline,” which includes: collection of tweets, feature extraction via Tweet2Vec, clustering to aggregate semantically similar tweets, ranking to extract characteristic tweets, and rendering of clusters of semantically related tweets through the visualization engine. The tool itself is “a desktop web application best viewed with Google Chrome” (p. 4). The authors tested the system on three datasets (Trump’s Immigration Speech, 2015; Orlando Shooting and Aftermath, 2016; Discussion of U.S. Economy, Summer 2016 to ascertain the narratives emerging from characteristic tweets. The authors concluded the tool provided a viable means to analyze large volumes of tweets in a short amount of time.  

‍