Discern commercializes search tool
Discern Communications has introduced a commercial version of software that enables natural language searches of structured and unstructured data
Discern Communications Inc., a 2-year-old spinoff from SRI International, has introduced a commercial version of software that enables natural language query-and-answer searches of both structured and unstructured data. It is targeting military, intelligence and civilian agencies as potential customers.
When used for automated, customer self-service applications in agencies such as the Social Security Administration and the Internal Revenue Service, Discern said organizations would see a return on their investment in the product in less than three months.
"Calls that come into agencies and that have to be handled by personnel there can cost dollars per call," said Steve Obsitnik, president and chief executive officer of Discern. "With our product, they can do it for cents per call, because many more calls can be handled outside of the agency call center."
Discern's product is the result of a 10-year SRI research project and a $25 million investment by the government begun when it discovered that military commanders during the Gulf War had no easy way to get cogent analyses from the mountains of digital information that were being collected. The situation has worsened since then.
"There's just an exploding amount of information being produced in multiple languages, but so far there've been few ways to get to that information," Obsitnik said. "And that applies now as much to nondefense organizations."
There is a huge demand for the kind of capabilities that Discern says its products can deliver, particularly with the ability to search both structured and unstructured data, according to Katrina Howell, an industry analyst with Frost and Sullivan.
"There are lots of challenges [for contact centers] in terms of the expense and complexities involved in dealing with unstructured data now," she said. "You basically have to have technicians come in and create a library from that unstructured data before you can search it. If you can [directly search] that unstructured data...you minimize that expense."
The contact center market is growing fast, she said, though it has slowed somewhat recently because organizations are unwilling to make any capital outlays in the current economy. But self-service is one area where the potential savings are so huge that organizations will definitely not forgo investments, though they may defer them, Howell said.
Discern's solution fits well with the current convergence of information sources, said Brett Shockley, CEO of Spanlink Communications Inc., a vendor of contact center solutions. Spanlink recently made an equity investment in Discern, and will help with the sales and marketing of Discern's product.
"The world of call center technology has broadened out into what is now known as customer contact technology, where contacts can be made via e-mail, through a Web page or with a phone call," Shockley said. "There's no sense in creating separate systems to do that, and in fact it's easier [to handle customer queries] when the whole thing is merged."
A major advantage of the Discern product, he said, is that it can link back to what is happening in the call center. It can provide an answer to a query itself, or if more help is needed, can direct a caller to the agency best able to answer the query.
Robinson is a freelance journalist based in Portland, Ore. He can be reached at hullite@mindspring.com.
***
Keeping it simple
Discern Communications Inc.'s Dynamic Context technology works by defining relationships among key words within a customer query to understand the intent and purpose of the inquiry. It also has a contextual understanding of the information sources throughout an enterprise and can therefore quickly map the query to the relevant data.
Company officials stress that the product is not a search engine. Instead of broadcasting a general query and getting back hundreds or thousands of possible answers or Web links that the user then has to sort through, Discern's product narrows requests and pulls back a more manageable set of answers for the user.
The requests can be made via a Web, e-mail or speech interface.
NEXT STORY: Addressing Postal questions