On the Way to a Science Intelligence: Visualizing TEL Tweets for Trend Detection

This paper presents an adaptable system for detecting trends based on the micro-blogging service Twitter, and sets out to explore to what extent such a tool can support researchers. Twitter has high uptake in the scientific community, but there is a need for a means of extracting the most important topics from a Twitter stream. There are too many tweets to read them all, and there is no organized way of keeping up with the backlog. Following the cues of visual analytics, we use visualizations to show both the temporal evolution of topics, and the relations between different topics. The Twitter Trend Detection was evaluated in the domain of Technology Enhanced Learning (TEL). The evaluation results indicate that our prototype supports trend detection but reveals the need for refined preprocessing, and further zooming and filtering facilities.

• Type: Journal
• Year: 2011
• Where: ECTEL
• Authors: Peter Kraker, Claudia Wagner, Fleur Jeanquartier, Stefanie N. Lindstaedt
• Keywords: ECTEL 2011, Machine Learning, Temporal Evolution, Tweets, Twitter