22
Predicting the Future With Social Media

Predicting the Future With Social Media. Introduction Goal – How buzz and attention is created for different movies and how that changes over time

Embed Size (px)

Citation preview

Predicting the Future With Social Media

Introduction• Goal– How buzz and attention is created for different movies and

how that changes over time

– How sentiments are created, how that propagates and how they influence people

• Hypothesis– Well-talked movies will be well-watched

Prior Work• Using meta-data information of movie

– Genre– MPAA rating– Running time– Release date– Number of screen– Actor– Director– Etc

• W. Zhang and S. Skiena. “Improving movie gross prediction through news analysis.” In Web Intelligence, 2009– Used a news aggregation model along with IMDB data

Dataset Characteristics• 2.89 million tweets / 24 different movies / 3 months

• Critical Period

re-lease

1 week

2 week

Dataset Characteristics

Dataset Characteristics

Dataset Characteristics

Dataset Characteristics

Attention and Popularity• Prior to the release of a movie

– Expect the tweets to consist promotional campaign– Tweets and retweets referring to particular urls(photo, trailers, …)

Attention and Popularity

Attention and Popularity• Correlation between urls and retweets with the box-office rev-

enues

• Tweet-rate

Attention and Popularity• Correlation between avg tweet-rate and BO revenues = 0.9

• Strong linear relationship => linear regression model

• Prediction of first weekend Box-office revenues

Attention and Popularity• Comparison with Hollywood Stock Exchange

Attention and Popularity• Prediction revenues for a given weekend– Using Tweet-rate timeseries + thcnt

Sentiment Analysis• Tweets are classified into Positive, Negative or Neu-

tral

Sentiment Analysis

Sentiment Analysis

Attention and Popularity• Prediction revenues for a second weekend

Conclusion• Social media feeds can be effective indicators of real-

world performance

• Tweet-rates can be used to build a powerful model for predicting movie box-office revenue

• Sentiment in tweets can improve box-office revenue prediction after the movies are released

END