What does Twitter look like? One way to look at it is as network of ideas and people that are linked by being mentioned in the same Tweet or by the same person. Below is a network of the top 300 current hashtags and user mentions on Twitter over the last twelve hours, with those in the last four hours weighted more heavily. Ideas or people mentioned frequently are show in larger circles, while those that are mentioned together are linked by lines and color. Hover over a circle to reveal the associated Twitter keyword or topic. This representation works best in Safari and Chrome.
Data is from the Twitter gardenhose stream which continuously sends out a random sample of approximately 1% of all tweets. This is collected in Python using Tweepy. A network of co-occurrences is then analyzed with NetworkX, and related nodes are clustered using the Louvain method for community detection. Node size is based on the square root of the number of mentions, with adjustments so that the total circle area is across graphs is comparable. Edge width is proportional to the square of the number of co-occurrence and edges with sample weights below a threshold of three are not displayed. We also remove the #porn and #followmeback communities. Data is exported using Drew Conway's NetworkX fork, and displayed using Mike Bostock's d3.js force layout.