Monthly Archives: July 2011

Twopical

Twopical is an interface to Twitter Discussions. It listens to tweets that contain a given keyword (or set of keywords) and it displays the topics that emerge from the discussion by visualizing them in real time.
The outcome of the visualization is a map of the semantic space of the Twitter discussion: topics that are usually cited together appear close to each other, popular topics increase their visual relevance in the representation, forgotten topics that are no longer part of the discussion slowly disappear, and old topics that have always been present in the visualization get more and more stable over time.

Objective of the interface is to give an overview of the current discussion taking place in twitter around a certain theme, in terms of recurrent words and affinity between these emerging topics. This experiment has been developed in the context of a research residency at IMéRA (Institut méditerranéen de recherches avancées) in Marseille (France), in the context of a research on the representation of dynamical networks. Starting from some reflections on the availability of digital traces of “conversations” in social media, the research team explored possible strategies for the representation of a conversation about about a given topic on the Twitter micro-blogging system. Several techniques were deployed to extract the meaning out of the noisy stream of Twitter data, and prototypes were developed with the goal of representing the discussion space and the activity of users in this space.

While a final version of the project could not be completed during the IMéRA residency, the prototype designed at IMèRA has later been further developed and it was eventually used to map the Twitter semantic landscape at the McLuhan Galaxy 2011 Conference in Barcelona (Spain). Currently, the concept is under further development in order to explore different directions in the representation of dynamic spaces defined by semantic networks and human interactions.

The Egyptian Revolution

On February 11 afternoon, André Panisson was testing a Python server that connects to Twitter and converts the stream of tweets and retweets in a format that can be read by the Gephi Graph Streaming plugin. In order to check if it was working, he used one of the most active hashtags at the moment, the #jan25 hashtag used by the Twitter community to refer to the Egypt Protests, and at some point there was a burst in the activity: Egypt’s vice-president had just made the resignation announcement.
In real time, this fortuitous visualization was displaying the exact moment in which Tahrir Square, from a mass protest demonstration, had transformed in a giant party. And simultaneously it was showing the virtual counterpart of the event, taking place in the Twitter network.

The video

The video shows the network of retweets containing the hashtag #jan25: if a person forwards (retweets) somebody else’s message, these two users will appear in the graph as nodes connected by an edge. The aggregation of such events displays how different users (the nodes) relay other user’s messages (the edges).
Before the announcement, the dynamics of the network are relatively calm — few people retweet each other — but when the announcement arrives, people start retweeting users and news sources from whom they got the information, and bursts of retweets appears on the network.

Tools and data

This visualization has been realized by André Panisson using:

The dataset is available in GEXF format.