Twitter is not just for tweeting anymore. With the access to Twitter’s functionality and data through it’s API, researchers and the general public alike have the ability to build their own applications to do just about anything. Take for instance, Martin Robbins, science blogger for the Guardian. He wanted to answer the loaded question… Do women science bloggers tweet differently to men?
In the old days, one would have had to create a screen scraping bot to view the timelines of all of the accounts in the study and save the pages to a database where the data could be tagged and analyzed. It would have taken a decent amount of time to program the bot to know what to save and what to toss. Mind you this is just the data collection step of the scientific method… all before you can get to the meat of actually investigating the hypothesis.
With your favorite scripting language and access to Twitter’s API, the task of assembling the data becomes almost trivial, leaving more time for the data analysis. As Robbins explains:
“Using the Twitter API, I grabbed the last 100 tweets from each person (representing on average just over 10 days’ worth), and indexed every word, mention, link, and hashtag in a database, along with various user statistics such as follower and friend counts, tweet frequency, retweets and so on.”
Robbins leads off with how unscientific his method was, as the dataset was small and probably skewed somewhat by his own bias, but the results are interesting. Take a look at the full article to see what he found.
Indigenous Tweets a new site from computational linguistics professor Kevin Scannell hopes to build niche online communities based around languages. A recent study by Scannell showed that there were almost 500 languages in use on Twitter. A few notable esoteric examples include languages such as Kreyòl ayisyen (spoken by some 12 million people in Haiti) in use by 6,878 Twitter accounts, Basque with 3,000 tweeters and Gamilaaray, an Australian indigenous language with three living speakers and only one who tweets.
The site is generated by using a program that “crawls” Twitter users, grabbing the tweets on their timeline and performing statistical language recognition on those tweets [...] Then, if a given user has more than a certain fraction of their tweets in the target language, that user’s followers are added to a queue to be checked in the same way. In the last couple of days, the initial crawls for Basque and Welsh were completed.
In Japan where power outages are still an issue since the recent earthquakes and tsunami, Naoki Hoshino, a test engineer at Peritec is using Twitter to provide communication between remote testing sites. He built an application to monitor battery levels and voltage at locations where they are conducting experiments. At regular intervals, his app hits the Twitter API to tweet the results so that he and fellow engineers have early warning when the batteries need replacing.
Do you have an interesting usage of Twitter’s API? Let us know in the comments below.