Even though computers can't read, they're very effective at extracting information from natural language text. They can determine the main themes in the text, figure out if the writers of the text have positive or negative feelings about what they've written, decide if two documents are similar, add labels to documents, and more.