UMD Startup Delivers Simple Solutions For 'Big Data' Challenges

UMD Startup Delivers Simple Solutions For 'Big Data' Challenges

Predicting elections and the behavior of people on social media are just two of the many offerings from SentiMetrix, a University of Maryland startup developing new technology to help understand people’s emotions and analyze language, opinions, and other metrics on social networks.

The company is hard at work providing solutions related to medicine, national security, and telecommunications, among others.

V. S. Subrahmanian, Professor of Computer Science at UMD, is a co-founder of the company and inventor of the technologies behind the products and services offered by SentiMetrix.

“The starting point was a ComputerWorld award given to the OASYS [Opinion Analysis SYStem] developed by my grad school adviser Dr. V.S. Subrahmanian,” said Vadim Kagan, co-founder and President of SentiMetrix. “It was immediately clear to me that building a company around such a technology was something I was very much interested in.”

SentiSocial, one of SentiMetrix’s products, summarizes opinions of influential users on social media platforms, providing valuable information to its clients on how they can modify their products and content.

SentiMetrix offers solutions by mining “big data,” social media analytics, and predictive modeling.

“In our case it is often data from social networks, blogs, and news – but it could include other data sources, from fitness tracker devices to internal corporate statistics,” Kagan explained. “Big Data involves huge data sets. The term itself tends to be used to characterize data sets so big that traditional processing methods no longer work—for example, imagine trying to work with an Excel spreadsheet with 500 million rows.”

SentiMetrix technology performs real-time analysis on huge data sets in many languages to determine solutions to various challenges when it comes to predicting user behavior.

Take SentiGrade for example, an analysis engine for customers who have their own analytical solutions but need to derive sentiment scores from their data.

“The simplest category is determining users' opinions toward a particular topic—brand, product, political figure, policy, etc.—how positive or negative are the opinions, how do they change over time, and what influences these opinions,” said Kagan.

Kagan said that developing technology to accurately track emotions was one of the challenges they faced. Answering questions like: “Do people dislike something because they are afraid of it or are angered by it?” was another challenge.

When users express emotions and opinions in the form of text, it can be used for a variety of purposes. Determining the overall mood of certain populations, detecting early signs of some psychological conditions, predicting violent events, are some of them, Kagan said.

 “Predictive analytics is a rapidly growing part of our services—it helps our clients to proactively plan for tomorrow, rather than being limited to just the reactive post-factum responses,” he said.

While SentiGrade measures intensity and polarity of opinions, SentiBility is focused on predicting stability of countries—a service useful for government agencies and large international corporations and organizations.

The company has changed a lot since its inception, Kagan said. “Currently most of our clients come from government agencies.”

“SentiMetrix has significantly expanded its capabilities since the early ‘pure sentiment’ days. We no longer focus exclusively on sentiment analysis, but have a much more comprehensive set of offerings, addressing today's Big Data needs of our clients,” he said.

The SentiMetrix client and partner list includes The World Bank, Stanford University, United States Army, Forbes, and Tata Consultancy.

The Office of Technology Commercialization guided the founders through the processes of licensing the original core sentiment analysis software.

Kagan envisions that in the next five years, SentiMetrix will grow into a leading analytics provider, joining the league of companies like ClaraBridge, Microstrategy, and IBM.

“It is hard to believe that SentiMetrix is already almost 9 years old. Time flies when you are having fun,” he said. He also hopes that more UMD students will get on board with working in the company.

“I believe College Park has a lot of exceptionally bright students and I hope to see more of them on our team,” he said.

To learn more about SentiMetrix, visit: www.sentimetrix.com.

November 9, 2015