Unstructured Data Analytics: Replacing ‘I Think’ With ‘We Know’

Anyone who reads our blogs is no doubt familiar with structured data—data that is neatly settled in a database. Row and column headers tell it where to go, the structure opens it to queries, graphic interfaces make it easy to visualize.  You’ve seen the resulting tables of numbers and/or words everywhere from business to government and scientific research.

The problem is all the unstructured data, which some research firms estimate could make up between 40 and 80 percent of all data.  This includes emails, voicemails, written documents, PowerPoint presentations, social media feeds, surveys, legal depositions, web pages, video, medical imaging, and other types of content.

Unstructured Data, Tell Me Something

Unstructured data doesn’t display its underlying patterns easily. Until recently, the only way to get a sense of a big stack of reports or open-ended survey responses was to read through them and hope your intuition picked up on common themes; you couldn’t simply query it. But over the past few years, advances in analytics and content management software have given us more power to interrogate unstructured content.

Now OpenText is bringing together powerful processing capacities from across its product lines to create a solution for unstructured data analytics that can give organizations a level of insight into their operations that they might not have imagined before.

Replacing Intuition with Analytics

The OpenText solution for unstructured data analytics has potential uses in nearly every department or industry. Wherever people are looking intuitively for patterns and trends in unstructured content, our solution can dramatically speed up and scale out their reach.  It can help replace “I feel like we’re seeing a pattern here…” with “The analytics tell us customers love new feature A but they’re finding new feature B really confusing; they wonder why we don’t offer potential feature C.”  Feel more confident in your judgment when the analytics back you up.

The Technology Under the Hood

This solution draws on OpenText’s deep experience in natural language processing and data visualization.  It’s scalable to handle terabytes of data and millions of users and devices. Open APIs, including JavaScript API (JSAPI) and REST, promote smooth integration with enterprise applications.  And it offers built-in integration with other OpenText solutions for content management, e-discovery, visualization, archiving, and more.

Here’s how it works:

  1. OpenText accesses and harvests data from any unstructured source, including written documents, spreadsheets, social media, email, PDFs, RSS feeds, CRM applications, and blogs.
  2. OpenText InfoFusion retrieves and processes raw data; extracts people, places, and topics; and then determines the overall sentiment.
  3. Visual summaries of the processed information are designed, developed, and deployed on OpenText Information Hub (iHub).
  4. Visuals are seamlessly embedded into the app using iHub’s JavaScript API.
  5. Users enjoy interactive analytic visualizations that allow them to reveal interesting facts and gain unique insights from the unstructured data sources.

Below are two common use cases we see for the OpenText solution for unstructured data analytics, but more come up every day, from retail and manufacturing to government and non profits.  If you think of further ways to use it, let us know in the comments below.

Use Case 1: On-Demand Web Chat

A bank we know told us recently how its customer service team over the past year or two had been making significantly more use of text-based customer support tools—in particular pop-up web chat.

This meant the customer service managers were now collecting significantly more “free text” on a wide range of customer support issues including new product inquiries, complaints, and requests for assistance. Reading through millions of lines of text was proving highly time-consuming, but ignoring them was not an option.

The bank’s customer service team understood that having the ability to analyze this data would help them spot and understand trends (say, interest in mortgage refinancing) or frequent issues (such as display problems with a mobile interface). Identifying gaps in offerings, common problems, or complaints regarding particular products could help them improve their overall customer experience and stay competitive.

Use Case 2: Analysis of Complaints Data

Another source of unstructured data is the notes customer service reps take while on the phone with customers. Many CRM systems offer users the ability to type in open-ended comments as an addition to the radio buttons, checklists, and other data structuring features for recording complaints, but they don’t offer built-in functionality to analyze this free-form text.  A number of banking representatives told us they considered this a major gap in their current analytics capabilities.

Typically, a bank’s CRM system will offer a “pick list” of already identified problems or topics that customer service reps can choose from, but such lists don’t always provide the level of insight a company needs about what’s making its customers unhappy.  Much of the detail was captured in unstructured free-text fields that they had no easy way to analyze.  If they could quickly identify recurring themes, the banks felt they could be more proactive about addressing problems.

Moreover, the banks wanted to analyze the overall emotional tone, or sentiment, of these customer case records and other free-form content sources, such as social media streams. Stand-alone tools for sentiment analysis do exist, but they are generally quite limited in scope or difficult to customize.  They wanted a tool that would easily integrate with their existing CRM system and combine its sentiment analysis with other, internally focused analytics and reporting functions—for example, to track changing consumer sentiment over time against sales or customer-service call volume.

A Huge, Beautiful Use Case: Election Tracker ‘16

These are just two of the many use cases for the OpenText solution for unstructured data analytics; we’ll discuss more in future blog posts. You may already be familiar with the first application powered by the solution: the Election Tracker for the 2016 presidential race.

The tracker, along with the interesting insights it sifts from thousands of articles about the campaign, has been winning headlines of its own. Expect to hear more about the Election Tracker ’16 as the campaign continues.

Meanwhile, if you have ideas on other ways to use our Unstructured Data Analytics solution in your organization, leave them in the comments section.

About Stannie Holt

Stannie Holt
Stannie Holt is a Marketing Content Writer at OpenText. She has over 20 years' experience as a journalist, market research analyst, and content marketing expert in the fields of enterprise business software, machine learning, e-discovery, and analytics.

Check Also

Find More Knowledge in Your Information at Enterprise World 2017

If your office is like most, it’s got millions of gigabytes full of information stashed …

Analytics Workshop

What a Difference a Day Makes: Get up to Speed on OpenText Analytics in 7 Hours

One of the biggest divides in the work world these days is between people with …

Leave a Reply

Your email address will not be published. Required fields are marked *