Poynter's Adam Hochberg has studied StatSheet, a small company that turns sports statistics into articles, just using computers. Having launched in November, StatSheet has created a network of 345 websites (so far), each devoted to a different US university's basketball team. The company's proprietary software takes the statistics and box scores and creates text about each game.

Hochberg spoke to StatSheet founder Robbie Allen, who is confident that there is a niche for the sites to fill. Although the automated writing sometimes leads to slightly odd phrasing and repeated clichés, it makes the raw statistics far more accessible to fans. And the software, with access to hundreds of past statistics, can sometimes draw out interesting facts that a human journalist might overook.

Allen told Hochberg that StatSheet's algorithm takes into account a team's record, the strength of its opponents and its momentum heading into each game. "We do a lot of different computations that will result in a specific type of sentence," he said.

"While we believe there will always be a role for expert and personality-based opinions and analysis, we believe a great deal of that knowledge and expertise can be codified. We are dedicated to codifying that knowledge and applying it to our proprietary database of statistics, so that interesting highlights, trends and analysis can be generated, curated and delivered in real-time," the StatSheet site says.

StatSheet, headquartered in Durham, North Carolina, and described by Allen on its site as a "labor of love," hopes to make money from selling ads, earning commissions from merchandise and ticket sales, and syndicating its content to other websites.

Undoubtedly, a computer is not going to be able to recreate the atmosphere of a sports game as a journalist could, and will not be able to accurately convey the emotion surrounding a game. And as Penn State Sports Journalism Professor Malcolm Moran said to Hochberg, the story of the day is not always about the statistics.

However, there is definitely an argument for using such technology on occasions when a journalist cannot be present: structured text does carry more meaning for most than bare statistics. And in a time when newsroom resources are precious, it is likely that more and more games will go uncovered. As Jason Fry, former Wall Street Journal columnist and editor, said to Hochberg, an advantage of making use of technology to cover straightforward stories means that journalists are freed up to work on more interesting pieces.

This is the same argument made by Owen Youngman at Northwestern University, whose Intelligent Information Lab has also produced automated journalism technology. The school's 'Authoring Engine' is used by the Big Ten Network, which focuses on college sports, and can "take away the rope work that stands in the way of doing thoughtful reporting," Youngman said earlier this month. He suggested that the technology could also be used for finance, politics, real estate, crime or hyperlocal: anything with a strong statistical element.

StatSheet intends to move into other sports soon. Its sites will all have their own Twitter account, Facebook fan page and mobile application, the website says.

Clearly, computer-generated writing cannot replace high-quality journalism by experienced reporters, and leaving any work up to computers will mean risking missing stories. But for stories which would otherwise go unreported, experimenting with this automated reporting could well be valuable. And even when there are journalists present, it could be worth making use of the data analysis this technology offers.

Source: Poynter