IBM believes the future of business intelligence (BI) involves unstructured information, real-time analytics and semantic search -- but the present revolves around partners not purchases. IBM's Information on Demand event this week had a special BI keynote speech and several relevant sessions, but despite Wall Street speculation that IBM would buy a BI vendor like Business Objects or Cognos, there were no big acquisition announcements. Karen Parrish, IBM's vice president of BI solutions, laid those rumors to rest during an interview at the event.
"In the area of traditional BI -- like query, reporting and OLAP -- we're going to partner. Our partners are mature, accepted by the market and good at what they do," Parrish said. "Many of our customers have already made investments in that space. For us to introduce something new or to upset the marketplace isn't going to benefit IBM and isn't going to benefit our customers."
On the exposition floor, IBM demonstrated several custom BI applications that incorporated many of the concepts touted by Parrish. In one session, attendees listened to the potential business cases for futuristic BI, laid out by Sreeram Balakrishnan, architect for discovery solutions with IBM. There are very real ways that next-generation BI can help companies solve problems and save money on customer care, market intelligence, product quality and research and discovery, he said.
IBM researchers have built applications leveraging Big Blue's unstructured information management architecture (UIMA), OmniFind search tools, WebSphere Content Discovery, deep analytics and advanced data visualization technologies. A key foundation is what IBM is calling an "unstructured enriched data warehouse," Balakrishnan said, which entails extracting information from unstructured data and appending it to structured data for analysis.
Futuristic BI in action
Balakrishnan offered a few demonstrations to put the concepts in context. Take for example a custom call center application that combined structured data about a call, such as the length of the call and the agent, with unstructured data extracted from call transcripts, emails and survey results. This enables companies to construct more sophisticated customer models, better assess lifetime value, better train agents and optimize self-help resources, which are activities that can lead to lower costs and increased profits, Balakrishnan said.
Another demonstration centered on a rental car company that was having problems with no-shows, people who made reservations but never picked up the car. After analyzing transcripts of agent communications and comparing them to car pick-up rates, they found that customers were more likely to show if the reservation call included many "value-selling phrases," like mention of a good rate or nice car. Eventually, the company plans to use speech analytics, rather than text transcripts, Balakrishnan said. Another company uses speech analytics already to identify out-of-the-ordinary calls, which are then tagged for a human to listen to, Balakrishnan said. It's enabled the company to zero in on just the calls that require human analysis, without sifting through uneventful calls.
Another futuristic IBM demonstration on the expo floor analyzed video footage and identified similar concepts in images. Demonstrators showed that a regular search on "soccer," based on the text captions, returned hundreds of results, such as anything on soccer stars or soccer moms. However, when they used advanced image analysis, they could produce a short list of footage showing actual soccer games.
In almost every demonstration, natural language search was the starting point -- an important concept that will make BI accessible to a broad user community. Search is getting better at understanding what people mean, not just what they say, Balakrishnan said. That means understanding that a search for "Barbara phone" does not want every document with those words in them; rather, it's a request for Barbara's phone number -- which could be stored in a structured or unstructured data source.
It's important for businesspeople and technologists to know what's possible now and what will be possible soon, said Paige Allen, director of strategic accounts with Atlanta-based Logical Business Solutions. She attended Balakrishnan's presentation so she could educate the consulting firm's project managers, who often work on projects involving clients' unstructured data.
"I want [our project managers] to know what's available so they can better visualize their options. They look at a project now, and might just look at structured data or the existing way that it's done. They need to be taking a better look at some of these newer innovations," Allen said.