April 23, 2001, 3:31 PM — Text, video, audio. It's all here, living and replicating like a corporate case of foot-and-mouth disease. The avalanche of information pays little heed to networks, operating systems or applications. Indeed, it's amazing you can find anything, let alone what you need when you need it.
Clearly, search technology is evolving into a set of knowledge management tools to let enterprise users locate information on databases, in e-mail repositories, across intranets and on the Internet. And those tools must interact with existing applications. For example, when you find the video and text files you want, you should be able to move them across a variety of networks and actually do something with them.
John R. Sack, associate publisher and director of Stanford University's HighWire Press, told me he needed a program flexible enough to recognize the similarity of concepts as well as words for the 245 specialized journals he publishes online. "We had to find a way to offer finer-grain results for publications such as the Journal of Biological Chemistry without adding to the burden of individual classification."
Of course, you can manually tag every piece of information as being potentially interesting, but that's not practical when you consider "there are more documents behind the corporate firewall than on the Internet," according to David C. Peterschmidt, president and CEO of Inktomi Corp.
A better way is to create taxonomies, or sets of relationships that have distinct meanings to users, such as determining certain noun phrases that are related. Next is to combine results with keyword searches; this creates what are called "latches," or connections to other keyword search results. At HighWire, Sack teamed up with Semio Corp. to automate the creation of taxonomies, produce tagging programs and detect when the latching process is and isn't working.
This improvement is still focused in discrete searches on a single network and platform. But consider the data in a large corporate bureaucracy: sales presentations, form letters, supplier data, training videos. The information might be more valuable if it could be connected and served to a user in a meaningful way. But if it exists across different networks and platforms, IT managers aren't going to shift terabits of data to accommodate a slick search technology.
In fact, it's worthwhile when getting search results to take into account where information lives -- such as on an intranet as opposed to the Internet -- as well as who created the data and when it was created.
Companies such as Inktomi are producing applications to do this. By overlaying the transport level (routers and switches), their new software can locate and then shift data around a network once you've found what you need.
It's time to extract the value out of your information by treating search technology as part of your IT infrastructure.