April 13 , 2010
Twitter chatter may predict movie success, study
"Tweet rate" may equal box office take
The rate of chatter on Twitter can be used to accurately predict how well a movie will perform at the box office, researchers at HP Labs have found.
The researchers examined 2.89 million tweets from 1.2 million users that related to 24 different movies released over a three-month period from November 2009 to February 2010.
They got their data by crawling hourly feed data from Twitter.com, using keywords present in the movie title as search terms.
The researchers wanted to know if the "tweet rate" — the number of tweets per hour — could be used to estimate the opening weekend box office take for movies before they opened.
According to their research, the results were incredibly predictive.
"There is a strong correlation between the amount of attention a given topic has [in this case a forthcoming movie] and its ranking in the future," the researchers wrote in their paper, Predicting the Future With Social Media.
For instance, the movie Transylvania generated just 2.75 tweets per hour in the week before its release. In its opening weekend, despite showing on more than 1,000 screens, the movie grossed just $263,941and was subsequently pulled from theatres in less than two weeks.
Avatar , on the other hand, generated 1,212.8 tweets per hour and pulled in $77 million on its opening.
The authors were able to design a model that provided accurate predictions of how much money the movies would bring in on their opening weekend.
Tweet sentiments predictive too
"We also analyzed the sentiments present in tweets and demonstrated their efficacy at improving predictions after a movie has [been] released."
For example, the researchers noticed a big uptick in the positive sentiment of tweets about the movie The Blind Side after its first week in theatres. The movie had a lukewarm first-weekend box office of $34 million, but improved to $40.1 million in the second week.
The movie New Moon, on the other hand, saw its box office plunge by more than two-thirds after its big opening weekend — an event predicted by a corresponding drop in the ratio of positive to negative tweets.
The authors of the paper said their analysis "can be extended to a large panoply of topics, ranging from the future rating of products to agenda setting and election outcomes.
"At a deeper level, this work shows how social media expresses a collective wisdom which, when properly tapped, can yield an extremely powerful and accurate indicator of future outcomes."
The unpublished paper is the work of HP Labs scientists Sitaram Asur and Bernardo Huberman.