Editor’s Note: I am super excited today to interview Rohini Srihari, Chief Scientist at SmartFocus for the Online MR blog. Rohini will discuss at Sentiment Analysis Symposium, (July 15-16, New York) Inferring Demographics from Social Data. In this interview Rohini provides in depth insights on various topics related to the insight industry.
Background of Rohini Srihari: She is an entrepreneur, scientist and educator. She has been involved with language technology companies for over a decade. She has worked extensively with the US Government, specifically the Department of Defense in developing innovative multilingual text mining solutions. Most recently, she founded Content Savvy, a company offering “big data” content enrichment and analytics solutions for various markets, including healthcare and customer service and support. Content Savvy was recently acquired by SmartFocus. Dr. Srihari is on the faculty of the Dept. of Computer Science and Engineering at the University at Buffalo. She has given numerous talks and published extensively in the areas of text mining and multilingual text analysis.
Akshay Kanyal: Can you briefly summarize your responsibilities and what challenges excites you in the current job?
Rohini Srihari: As Chief Scientist at SmartFocus, I am responsible for the identification of new technologies that extend our Message Cloud personalization solutions and add new value for our customers.
I am particularly focused on technologies related to natural language processing, machine learning, social media analytics and recommendation systems. Apart from the technology challenges, I am very excited to be able to work with customers and understand their needs.
It is most satisfying to be able to deliver solutions that demonstrate real ROI to our customers.
Akshay Kanyal: It seems that insight experts love to flirt with new methodologies but don’t seem to let go traditional approaches – is it feasible to sustain without embracing change?
Rohini Srihari: I think there is scope for both approaches, namely,
(i) a slow but steady series of incremental changes to improve existing technologies or methodologies and
(ii) a complete overhaul of an existing approach.
For example, speech recognition is where it is today due to decades of slow but steady improvements. On the other hand, new technologies such as deep learning are proving to be effective in challenging tasks such as image and video analysis.
And of course practically every technology, from search to recommendation systems have benefited through the use of machine learning.
The important takeaway is not to pursue change just for the sake of trying something new: one has to carefully evaluate the need for a totally new solution and the demonstrated benefits that it offers.
Often, the latest “shiny new thing” has only been tested in a research or lab environment and may not stand up to a production environment.
Akshay Kanyal: “Water, water, everywhere, not any drop to drink” is this rime of the ancient mariners applicable to the huge data that researchers deal with to extract insights?
Rohini Srihari: Well, this saying could be true if vendors simply deliver flashy analytics dashboards without consideration as to how these analytics could be leveraged by customers for some practical ROI.
In other words, the analytics need to be “actionable”.
I have heard customers express their frustration with big data analytics in terms of not knowing how to act upon the information, or telling them only things they already know.
Akshay Kanyal: With so many methodologies at our disposals is it becoming “too many cooks spoil the broth”? How should we go about selecting the right methodology for our research?
Rohini Srihari: The important thing is to focus on a valid and viable methodology, and there could be several.
A methodology, at least in the context of big data analytics, incorporates everything from content source selection, to accurate detection of various signals, to analytics that properly convey the impact of some trend.
The analytics need to be calibrated so that they account for population biases, and sensitive enough that emerging trends are picked up and not drowned out by chatter around consistently popular brands. Ideally, the analytics would also be bench-marked against some hard data, for example, past campaign data. In this way, the analytics can be used in a predictive manner, rather than as simply a post mortem analysis.
I think documenting a validated methodology is one of the key successes to good data science and increases customer confidence in the output.
Akshay Kanyal: According to you what seems to be the future of market research – especially in context of social media?
Rohini Srihari: Customers are asking more of social media analysis, especially in the context of understanding customer behavior.
In some cases, customers are interested in monitoring social activity of their current customers to identify any possibilities of “churn”. In other cases, customers want to learn more about their customers likes and preferences, so that they can offer more personalized services. And of course, customers are also interested in social media analysis for sales lead generation.
There are technology solutions for these, however many privacy issues must be addressed as well. I believe that discovery is an important area that we are just getting to: discovery of trends, product feature wish lists, service issues, etc.
Akshay Kanyal: What key points you want to convey through your presentation at Sentiment Analysis Symposium?
Rohini Srihari: I will focus on the benefits of understanding your customers – and how this translates into demonstrable ROI for businesses.
I will specifically discuss how you can infer demographical information from various sources including social media, and purchase histories. It’s interesting to see the kinds of “clues” that are useful, and how they can be combined to learn more about customers.
Akshay Kanyal: If we want to know more about your company what will be the best source of information?
Rohini Srihari: There is a lot of information available on our website, including data sheets for our Message Cloud solutions.
Our executives have given several interviews and written thought leadership pieces in several online publications. And of course, our marketing team is always willing to speak to anyone interested about our company and offerings!
Akshay Kanyal: What is the most important ingredient to become an awesome insight expert?
Rohini Srihari: Being an expert involves both a thorough understanding of what technology can and cannot do, as well as a firm grasp of how customers expect to use and evaluate technology solutions.
The best technology is the one that you don’t actually see, but where you feel the impact. It is also important to be able to properly attribute key performance gains to specific technology initiatives.
You can register with Sentiment Analysis Symposium HERE
You can save 10% by using ONLINEMR promo code while registering for the event!
Akshay Kanyal writes survey research reviews on his popular blog Online MR. He’s an avid blogger, brand consultant and a content marketing expert, helping business owners to craft content that sells.
He provides content marketing advice to start-ups and innovation driven companies. He can be contacted at firstname.lastname@example.org