Sentiment Analysis with Big Data
Comments, conversations, and customer feedback fill the Big Data universe. Through sentiment analysis, businesses are turning this unstructured data into actionable intelligence.
The ultimate dream of many Big Data users is to tame the vast frontier of unstructured data in social media, harnessing how people are thinking, talking, and feeling about a brand. Sentiment analysis scans tweets, pins, likes, and more for evidence of good, bad, or even indifferent impressions.
“Discovering what your followers are saying about you, from the conversations they have to the keywords they use, is crucial if you’re looking to pick up on positive and negative vibes associated with your brand,” website SocialTimes stated in August. “If you concentrate on your followers, you’re getting a select pool of data that allows you to focus solely on the comments made by those who matter.”
Bigger than Twitter and Facebook
But sentiment analysis is bigger than Twitter, Facebook, and the rest of social media’s domain, as myriad professionals try to analyze massive amounts of internal customer data too. That data is often outdated, incomplete, or stored in disparate systems, so those workers rely on instinct alone to make strategic decisions.
Businesses can unlock sentiment and contact insights from internal, company-owned sources, as well as social media with a new SAP HANA-powered application. SAP recently announced its Social Contact Intelligence, which allows organizations to:
- Use real-time sentiment and contact insights from social media and internal sources
- Identify and target key influencers
- Generate new leads and opportunities
- Improve overall service levels and customer loyalty
The app can help marketers drive incremental sales by incorporating social sentiment insights directly into their campaigns. Sales people can identify an account’s most influential and relevant contacts more quickly. And customer service professionals can analyze social channels for signs of trouble – and take preventive action.
Sentiment analysis still in its infancy
Sentiment analysis isn’t a panacea yet. It is still in its infancy, and there are limitations. “Despite significant advances in machine learning, it’s extremely difficult (or not practically efficient) for computers to understand and process natural language, automate sentiment analysis, or determine ambiguous context,” Wired stated just last month. “No matter how smart or stupid computers may be, it’s just easier to create systems that encourage users to do their own interpretative work.”
But the technology is evolving across many industries, including high frequency trading, which relies on algorithms to execute market orders – and make profits – within miniscule fractions of a second. Some firms are dabbling in media analysis for opportunities to trade on good or bad news.
Using structured data in trading, risk management, and fraud detection
“Conversion of unstructured data into an expanding universe of structured data will soon become ubiquitous,” Paul Rowady, senior research analyst at capital markets watcher TABB Group, wrote last month. “Trading strategies, risk analytics, fraud detection, and all sorts of decision support in global capital markets will eventually incorporate this converted data.”
In short, people are finding ways around the limits of sentiment analysis, as we would expect from an innovative society. So the dream of harnessing unstructured data on social media, company-owned systems, and even headline news is closer to reality than ever.
VIDEO: SAP’s new app is part of Customer Engagement Intelligence, which combines front- and back-end real-time customer information to form a 360-degree view of customers.