AIF Blog

Democratizing Data: Machine-generated Stories of Your Life

Jun 28, 2014

The machine age has given us a glorious gift.  We now have data about nearly everything in the world.  Businesses get masses of sales, logistics and performance data on a regular basis.  Individuals get data collected by their cars, thermostats and even health monitors like FitBit hooked to their bodies.  And, from the world of social, we have an unending flow of posts, tweets, videos and blogs that theoretically can be used to paint a picture of our world.


We tend to operate under the assumption that we can use this data to make better, smarter decisions about our lives.  In particular, we keep thinking that we can use the enormous computational resources available to us to look at all this data and perform clever analytics against it to provide us with a condensed version of the world, typically rendered in some type of table, chart, or graph.

But this assumption has two rather serious flaws. First, people have an astoundingly hard time understanding structured data in nearly all of its forms.  Second, machines have an equally difficult time understanding the world of unstructured data (otherwise known as words).  Essentially, we have a hard time talking to machines, and they have a hard time communicating with us.

The first problem plays out in the form of “big data” analysis that is genuinely incomprehensible to anyone except experts.  These experts created them in the first place; so the dream of making better decisions using these analyses only comes true for the data scientist.

Narrative Science has been successfully addressing this first problem with the development of Quill™, an AI system that is able to take structured data and use it to write highly informative and instructive narratives that transform the world of numbers into the world of insightful stories.  The system, which combines the skill sets of a data scientist with those of a journalist, is able to automatically craft narratives about the world through the lens of the data.

While Quill is a powerful tool, it is a machine, and so is itself prone to the second problem.  It does not have the ability to look at raw text and immediately figure out what it means.  Like other machines, it struggles with the unstructured world of social media and is simply unable to extract meaning from it.

Until today.

At Narrative Science, we've been working on a pilot project that addresses this second problem as well.  The project, Quill Connect, takes unstructured text drawn from Twitter, analyzes tweet streams as to topic and sentiment, and then produces a natural language report in plain English.  In particular, it analyzes the data from your Twitter behavior and crafts a story about your influence related to your followers and makes recommendations as to how you might extend your reach.  You then receive your personal story via Direct Message on Twitter.

You can try it right now at  Simply sign in and two minutes later you’ll have your story.  Not data, and not a chart, but a story about you and your followers with some insight and advice.


Kristian Hammond is chief scientist at Narrative Science, where he works on technology that generates natural language narratives based on the meaning and insight found in structured data. He will be a speaker on the session Everyone Has a Story: Mind-Blowing Technologies that Bring Narratives to Life, July 2.

Related speakers: