Tools capture unstructured service data

For most manufacturers, closing the loop on product quality for items that have left the plant is labor-intensive. To analyze even a representative sampling of customer comments—in the form of phone conversations, letters, or e-mails—employees usually must read and tally the data to get an idea of whether the available products are making customers happy. Edmunds.com, the Web site of the popular automotive buying guide, is "structuring" this kind of data using a tool from Attensity, such that it can be analyzed more thoroughly by business intelligence (BI) tools.

"We're taking data that our online forum moderators gather and we're feeding it back to our automotive industry customers," says Kerry Kim, manager of syndicated applications for Edmunds.com.

The challenge for automakers and most enterprises is that relational databases never capture most of this type data. According to Stamford, Conn.-based Gartner, more than 75 percent of all data is typically unstructured, leaving it outside the reach of BI tools. New data-mining techniques collect unstructured data and use sophisticated algorithms and logic to parse the information and convert it to a structured form that allows quantification.

Two companies—Attensity and ClearForest—have developed competing approaches that add structure to unstructured data. In essence, they look for the number of times specific search terms appear, analyze where the terms appear in a sentence or phrase, and then tally the number of occurrences to provide statistical data that can be analyzed.

Both providers also recently released vertical-industry modules that add industry context to their toolkits.

For instance, Attensity's Relational Extraction Server for the automotive industry looks for phrases such as "the fuel filter was plugged with black plastic shavings." For Edmunds.com, the Attensity system will analyze comments submitted to the site's online Town Hall product chat rooms.

ClearForest is releasing Quality Early Warning, which targets automotive warranty servicing—an area that the Automotive Industries Action Group estimates consumes somewhere between 1 percent and 3 percent of automotive industry revenues. "With our system, you could make the logical connection that dap conditions and fuel pump failures are related," explains Randy Clark, senior VP for ClearForest.