Tweet This
By Tom Connor
Statistics and facts are seldom enough. Businesses today want to know how customers feel in hopes of predicting their future actions, and they're increasingly turning to machines to achieve this deeper understanding.
Fueling these machines are text analytics algorithms that look at text and other unstructured data to determine how people feel about companies, specific products and services. This kind of data includes email, text messages and social media posts — essentially any information that doesn't reside in a spreadsheet or database.
Insights gleaned from unstructured data allow companies to understand and manage content they disseminate in new ways. Corporate assets such as email, SMS, blogs, survey responses and/or document archives become sources of data for analysis, and increasingly, they're also influencing business decisions.
Decades In The Making
While text analytics occupies the modern day intersection of artificial intelligence and computational linguistics, its roots trace back decades.
Researchers and developers began using text analytics predecessors in the 1950s to identify speech patterns, concepts and names in handwritten letters, typed reports and phone messages.
By the late 1980s, innovators developed the first machine-learning algorithms, and by 2000, businesses were using text analytics to draw insights from new kinds of digital material.
“When social media came in, we gained access to a vast amount of data where people were expressing opinions and sentiments," said Steve Pettigrew, senior technical product manager at OpenText, a provider of enterprise information management software. “Then, suddenly, we had a lot of value to leverage."
Single Platform Consolidates New Data Technology
The sheer volume of information generated in business today — a veritable tsunami of raw data streaming daily from ever-widening information channels — demands cutting-edge software that can work quickly to observe and analyze sentiment.
Some companies are using OpenText InfoFusion to perform those tasks. The platform, which replaces one-off information applications, works by extracting relevant linguistic elements from content, such as keywords, names and subject categories. The software is then able to determine the objectivity or subjectivity of content , and assess whether information is factual or emotional. If emotional, the software can tell whether it is positive or negative.
Once the sentiment from content has been measured, companies can use the information to track consumer tastes and trends, and predict responses to new product launches or services. Advances in visualization tools allow users to illustrate that data in charts and other graphic forms.
“Marketing analysts who want to analyze social content and understand customer sentiment toward their products and services will be able to visualize positive or negative trends over history and in real time," said Mark Gamble, senior director of technical marketing at OpenText Analytics.
To demonstrate these capabilities, OpenText is making a brief foray into the world of election data and presidential politics. Its Election Tracker '16 web app is crawling media coverage of the race to discern sentiment and trends that might go unnoticed by campaigns and pundits studying only structured data.
Text Analytics For The NBA Improves League Website
This past February, the National Basketball Association announced plans to use text analytics to analyze questions asked by fans visiting its NBA.com website. Rather than having to drill down through layers of official stats and charts, basketball fans will be able to compare, say, Stephen Curry and LeBron James, with greater ease and speed.
The U.S. Department of the Interior, an OpenText customer, uses text analytics to sort, categorize and prioritize the 2.7 million emails it receives each day. Using a transparent machine learning algorithm, text analytics allows the department to “drill down in the system and help build a better records management experience," Pettigrew said.
As enterprises seek a better understanding of the world around them and the data that increasingly defines it , wider use of these algorithms seems assured.
Tom Connor is a custom content business writer specializing in technology, entrepreneurism, health and sustainability.
