Search oughta Be in Pictures
Of course, text analytics only goes so far when an increasing amount of corporate data resources resides in image or video formats. The question of how to apply structure to these and other emerging formats is a growing challenge for enterprises. Kevin Smith, VP of sales for North America for LTU Technologies, a provider of image search solutions, says, “Enterprises are being forced to figure this out.” He cites pharmaceutical companies, whose key information assets may be stored in flow diagrams, as an example of an industry that must include image and video handling in its BI/search solution.
LTU got its start facilitating digital forensic analysis in law enforcement and intelligence, primarily for identifying stolen art, child exploitation, and counterfeit operations within a data environment. Increasingly, says Smith, “We are working with user-generated content sites to filter out or identify non-permissible content, or to locate instances of copyright infringement” on a near real-time basis. In another application, for stock photography agency Corbis, a user can type in a keyword phrase like “New York City” and pull back images with that tag. From there, users can drill down based on visual features, like colors or figures, rather than on keywords.
Smith says that “a combined text and image solution is a winning combination,” and he mentions that they are in early conversations with BI vendors to incorporate the visual angle into their search solutions.
More Access Has Risks
One potential downside of all this increased discoverability of data is the risk of people seeing information they aren’t meant to. Besides the headache of facing a disgruntled employee who has accessed the corporate database to learn that he is paid less than a colleague, there are larger and more serious risks.
LTU’s Smith mentions the proliferation of mobile devices to which files can be downloaded easily. “Especially as memory capacity of those devices increase, corporations have to be sure that they can secure their information assets. How do you identify a situation where an image has been downloaded to a device that can then be slipped in a pocket and walked out the door?”
Siderean’s Allen agrees that the increased ability to discover and blend information that was traditionally firewalled off raises risk. But he notes, “Vendors can model user roles into their solutions.” Both Semantra and Siderean’s products build strict security algorithms into their platforms, personalizing access based on the type of user. As corporations assess their own combined BI/search environments, they should keep a critical eye on security, and plan to revisit the topic frequently as formats and channels for accessing the data evolve.
The Future is Now
From the customer’s standpoint, the alliance between BI and search has already taken place. In their 2007 report “Don’t Wait for Search and BI to Become the Same Product,” Gartner’s Gassman and co-author Whit Andrews found that 90% of Global 2000 organizations expect search and BI to interoperate by 2012. That doesn’t mean, however, that vendors will present a single code stream just yet.
“The convergence will continue to move slowly,” says Gassman. “BI is still an expensive solution, so it’s not going to trickle down into the middle market right away,” he notes. The Gartner study concludes that search vendors and new market entrants may nibble at the edges of the traditional BI market by adding more analytical capabilities, but they will likely not displace those established solutions for process-driven or conventional BI capabilities. Conversely, BI vendors will continue to innovate and add on search capabilities, but they aren’t poised to make a major run for the search market.
Allen is enthusiastic about what he terms “a near-Cambrian explosion” of ways to visualize information that will drive the continued convergence. “There will be more mashups at the web level, which will speed the proliferation of search across BI applications,” in the near term. Both Allen and Davis believe that semantic modeling will be a major driver as well, providing a bridge between data and content in the BI world.
Wise mentions that as far as implementing search technology goes, BI was a relatively late adopter. “Search is still fairly new for BI vendors,” she notes. “And it’s only going to move forward.”
Companies Featured in This Article
Technology Evaluation Centers (TEC) www.technologyevaluation.com
Fast Information Search and Transfer (FAST) www.fastsearch.com
Google Enterprise Search www.google.com/enterprise/gsa/onebox.html
Information Builders www.informationbuilders.com
Business Objects www.businessobjects.com
LTU Technologies www.ltutech.com