Director of Data Science, Elias Ponvert Presents at Austin Machine Learning Group

The world of data science can be intimidating for anyone out of the loop. And if the name “Data Science” weren’t daunting enough, strange charts, symbols, and a lingo foreign to most have made the field easy to write off and leave to the “experts”.

But as Big Data becomes a buzz phrase, that attitude is becoming increasingly less common. Folks ranging from sociologists to CFOs, entrepreneurs to marketers are starting to pay attention to what the data has to say.

On Monday night, Director of Data Science at People Pattern Elias Ponvert gave a presentation of his thesis to a full house at the Austin Machine Learning Group at BlackLocus’ downtown digs. During the nearly 90 minute presentation, Elias shared his discoveries about unsupervised parsing to a diverse crowd in downtown Austin.

Hold on. What’s unsupervised parsing? Unsupervised parsing is automating the discovery of grammatical structure of sentences without any help from real people. Although considered a tough problem to crack, Elias was able to come up with a solution that outperformed the state-of-the-art method using what he explained as “fairly simple machine learning and hacks”.

Throughout the presentation, between Elias’ own brand of humor and dry-erase models and language trees, Elias walked his audience through his methods, pausing to hash out questions as they arose. In the end, folks left understanding a little more about the science behind languages and computers.

So what does it matter? For those in the marketing world, the concept of unsupervised parsing may seem completely foreign. As more information becomes available online, the problem of identifying what is said, and what is meant, becomes increasingly more important to identifying and activating valuable market segments and consumers. Unsupervised methods, like Elias’, can contribute to advanced text analytics in a wide variety of languages.

After the evening wrapped up, folks of all walks of life piled into a small, old-fashioned, elevator and for a short ride to the ground level, swapped sparks from across the industry board.

To read the dissertation and learn more about Elias’ work, please visit his personal webpage, or see his work in action with a demo of the People Pattern platform.