/// ‘Thirst’ Finds All the News That’s Fit for You
Most everyone who reads the news falls into one of two categories: those who know exactly which sources they like, and those who just want to read about the topics that interest them. Those who are loyal to their favorite blogs and websites have plenty of RSS apps to chose from for tablets and smartphones, the most popular of which are Flipboard and Pulse. But if you’re inclined to browse topics, skipping from one news source to another, there aren’t many apps to help you. San Francisco-based startup Thirst is filling that gap by using natural language processing to find articles that match your interests, even if they change constantly.
Though using natural language processing to find articles for you is nothing new (apps like Zite do it), Thirst CEO Anuj Verma believes he’s crafted a much more precise algorithm to find the most relevant pieces of writing. “Current natural language processors look at long-form articles with typically formal writing and can tell you roughly what the article is about,” says Verma. But we live in a world where news is delivered increasingly by Twitter. So Verma set out to build a language processor that could ascertain the meaning and context of a piece of writing with far fewer characters — even a 140-characters-or-less tweet.
Verma reasoned that if his language processor could understand tweets, with all the abbreviations and less-than-stellar grammar, it could do a much better job of looking at any piece of writing, and more precisely ferreting out the news and information that’s most interesting to an individual reader’s tastes.
Verma’s example is President Obama, who has many names associated with him. “He could be called POTUS, Obama, or a number of misspellings and hashtags that are meant to refer to him,” he says. “There are so many ways to refer to one person, and these older engines that were trying to classify things were missing how we actually refer to people today.”
The first incarnation of Thirst combed your Twitter feed and sifted out the most popular topics and tweets so you didn’t have to read everything in your stream. But when Twitter tightened its grip on its APIs in 2012, Verma got out of the Twitter app business, and turned his attention to general news. He figured that because he developed his natural language processor to parse short strings of text, it could pull far more of the context from a longer piece of writing, like a blog post or article.
Using the iPad app (also available for the iPhone), I’m inclined to compare it to Twitter’s trending topics, but with more context. Take a topic like the Oscars: there was much more being written about than just who won the big awards – some people talked about the host and others zeroed in on Michelle Obama’s cameo. If you opened the Oscars section in Thirst, you saw news that covered all of those angles and then some. As you use the app, Thirst learns what you like to read – and what you pass over – to show more articles that appeal to you.
For Verma, Thirst is just one step in his grander vision for how we’ll get our news in the future. “I prefer the NBA over the NFL, or the NHL or golf. When I go to Sport’s Illustrated’s website, why do I have to see the same thing that you see?” he says. He believes that everywhere we go, the content we see should adapt to our interests. That could mean one day the Wired.com homepage would look different for every person visiting it, with only the articles you might want to read and ads you might likely engage with.
Like most consumer startups these days, Thirst doesn’t have a business model yet, but Verma has ideas for how he could make money. One idea is to sell information to content producers about what you like to read. Verma knows not everyone will be comfortable with that, but he believes that ultimately most people will warm to the idea if they no longer have to wade through articles they don’t want to read to find the stuff they do.
Austin-based art and technology conference South by Southwest will be a huge test for Thirst to see if the technology can sift through the noise to find the most relevant content, and the different angles of the event. Since there’s news to be reported on the tech, art, comedy, film, and gaming fronts, Thirst will have to sort through each topic to help readers find the right stories. “SXSW will be a test of how Thirst works in an environment that has so many different angles of news,” says Verma. We’ll see if his system is ready for it.