Online Program

Firehose Data Science: Real-Time Analytics of Twitter Feeds (303574)

 View Presentation

The application of data science methods to social media feeds presents many challenges, including the creation of APIs to access the input stream, working with big data - especially high volume and velocity, and real-time decisioning. This paper demonstrates accessing Twitter feeds using a Java API, applying machine learning methods in SAS and R to classify tweets, and using these data to support further statistical analysis. The example used in the presentation involves mining the Twitter "firehose" for hate speech with a decision tree for classification, plotting and mapping the time series in real-time, and investigating the hypothesized connection between hate speech and acts of violence against persons targeted in the speech.