3 ms·

Seriously, What statistics would you get when to take in to the fact that 90% of the tweets are by bots? And twitter users with large following are either bots

Seriously, What statistics would you get when to take in to the fact that 90% of the tweets are by bots? And twitter users with large following are either bots or celebrities. All these factors should be considered while building the web app. Anyways it should be an interesting social experiment. Also, they should consider using facebook statuses instead of tweets. Facebook status messages are more personal than tweets. Any thoughts?

mpellon 17y ago
Facebook is being investigated as a potential data source as well especially given the acquisition of FriendFeed and roll out of their new search engine. I am of the researchers involved in the project (full disclosure) and yes we are concerned about the numbers of tweets generated by "bots" or other software although we estimated to be much lower ~25% (still significant mind you). We receive approx. 1.8 million tweets a day but it is important to note unlike similar projects we make no attempt to quantify the sentiment of individual tweets and instead only say something about significantly large portions of them (hourly, daily, weekly, monthly) time frames. Our method has proven useful on song lyrics, presidential speeches and blogs. See our publication in the Journal of Happiness Studies (http://www.springerlink.com/content/757723154j4w726k/?p=b9b0726c4baa4c55a845dec88c63f4b4&pi=2 http://www.springerlink.com/content/757723154j4w726k/?p=b9b0...) for more info, open access license.