i202 Fall 2010 School of Information, UC Berkeley

StatSheet recently launched a network of websites devoted to individual college basketball teams. All the content and articles on these individual sites are apparently written by software not humans. The founder of StatSheet, Robbie Allen, refers to these sites as "Robot Army" and this type of writing activity as "robo journalism."

I read this one story about Ohio State and it sounded pretty ok for a story written entirely by software, especially since the story-writing software doesn't even perform any linguistic analysis. It just uses a template of sentences and a database of phrases to compose a short article. It is a very simplistically written story, but I am somewhat impressed by some of the article's choice of words, such as

"The game lacked a lot of drama, with Ohio State up 52-25 at halftime and never letting up" 

The software is also suppose to be smart enough to write an article that is sympathetic toward a particular team. So one story written on a particular team site will have a different "sentiment" vs. on the opposing team's site, even though both stories are talking about the same game. Here's the counter part of the Ohio State article on the North Carolina team site, which is the team who lost that game.

Here's the NYTimes article that prompted me to checkout StatSheet's sites.