Comcast uses technology from CIA-backed CallMiner to mine data from every recorded customer service call in its Midwest call center.
(www.gridcomputingplanet.com Via Thomson Dialog NewsEdge)
Comcast is using grid technology from a CIA-funded startup to analyze recorded calls in its Midwest call center.
CallMiner's Virtual Server Room (VSR) distributed processing platform lets Comcast analyze every recorded call in the center. VSR makes medium and large speech analytics installations easier and cheaper by using a grid-based indexing platform to speed analysis. The company boasts funding from CIA-backed In-Q-Tel, among others.
Comcast's 200-seat Midwest call center generates thousands of hours of recorded calls each day. A typical speech analytics solution would take 20 or more servers to process Comcast's audio into data, but CallMiner's VSR leverages the existing unused computer processing power of the call center's PCs and servers to process all recorded calls with the startup's Analytics Suite. Comcast can now analyze all recorded calls without having to invest in additional hardware to process all the call data.
"Previously, we were not able to effectively listen to the majority of our recorded calls, and we knew there was valuable information that we were missing," said Joe Elias, the call center's technology manager. "CallMiner's VSR allows us to maximize our investment in call recording equipment and be proactive in our customer service approach by identifying early on how we can improve customer satisfaction."
CallMiner was founded in 2002 by Jeff Gallino, CEO, and Cliff LaCoursier, senior vice president of business development and marketing, both veterans of Thinkengine Networks, a speech recognition hardware manufacturer.
The CallMiner Suite includes CoreEngine, TrendMiner and Tireless Supervisor. CoreEngine converts recorded calls into searchable data. CoreEngine uses advanced statistical methods that analyze the audio with acoustic and language modeling to determine text from an audio stream and then statistically classify the speech recognition results for later analysis.
TrendMiner is a Web-based trending and analysis application that allows users to conduct queries and trend analysis on recorded calls and convert them into text and data, which is then stored in a relational database for data mining, retrieval, analysis mining and reporting. Tireless Supervisor automates call scoring and classification, giving users insight into operations and business processes.
The 451 Group analyst Aidan Biggins sees promise in CallMiner's technology.
"CallMiner promises to help call centers evolve into genuinely useful sources of customer data that's important to business areas like marketing and product development," Biggins wrote in a report last month.
He added that "the ability to process speech into useable data has broader potential. Other audio sources can be just as easily processed, allowing people to search for video, podcasts, or even AdSense-style, context-sensitive ads on Skype. Of course, this is based on the assumption that CallMiner's voice recognition software is up to the task."
One limitation, Biggins noted, is that VSR "can only batch-process voice recordings, making impossible any chance of automating CRM/BI responses in real time."
"It's a compelling story, yet CallMiner must move fast before companies with greater resources namely IBM leap in," he concluded.
Back To Grid Computing Planet
Copyright 2003 Jupitermedia Corp. All rights reserved.
Republication and redistribution of Jupitermeida Corp. content is
Expressly prohibited without the prior written consent of Jupitermedia
Corp.. Jupitermedia Corp., shall not be liable for any errors
or delays in the Content, or for any actions taken in reliance thereon.
[ Back To TMCnet.com's Homepage ]