First Concept
First concept based on data mining of Twitter > OccupyWallStreet tweets
Each tweet is analized with Pattern languange library to understand what's the polarity of each tweet, maybe in relationship to some specific event like people arrested in raids
UPDATE:
Data-mining with Pattern
search tweets talking about #ows
between november 15th and 17th
organized by date, hour of the tweets
filtered by text "arrested" or related concepts
1.577 tweets
Final csv file resorted with PHP script to produce a normalized timestamp in millisecs
Tweets are ordered by timeline from oldest to newer, on vertical axis
The polarity of each tweet (i.e. if each tweet expresses a bad or good opinion) defines the position on horizontal axis. The bad ones on the left, the good ones on the right. Tweets in the middle are neutral.
The size and the trasparency of each tweet is related to the number of retweets it had.
