Atypical web search can be time-consuming and ineffectual. To get to the information you actually need, you have to know the exact words in order to bring up the right results (or wade through masses of results). If you only have a vague idea of the document you are looking for or if you misspell a name, you may be out of luck at finding anything. Now, bring this common search experience into the enterprise and the stakes—and frustration levels—are even higher.
Dynamic navigation provides an approach to search that strives to change this aspect of search frustration. Instead of assuming that all users will take the same path to get to an answer, dynamic navigation provides a variety of routes to information, often in ways that can change on-the-fly.
The fact that navigation is part of the term dynamic navigation in itself indicates an important change for the search market, according to Susan Feldman, research VP of the content technologies group at IDC.
"In enterprise, finding the right answer has a lot depending on it, so you want them to find the right information without overwhelming them," says Feldman. "Any time you say navigation, it means browsing, which is quite different from search. Searching uses keywords that don’t often match up between the expert defining them and the layman searching. Browsing is another way of getting information by narrowing down categories along a prescribed path so people can make choices."
"Dynamic navigation offers a way to refine your request in the process of doing a search," says Jerome Pesenti, chief scientist at Vivísimo. "It serves two purposes: One is that it lets you find what you want by refining your search through choices, and the second is it gives you an overview of the content. It’s not just about drilling down the content, but also seeing what content is out there."
Mike Moran, distinguished engineer, content discovery at IBM, says the term dynamic navigation also describes any technique that allows a website to present information based on business rules rather than on manually architected link structure. "Anything that personalizes a website, anything that uses a search back end, all those kinds of technologies can be used to create a dynamically navigated site," says Moran. "Often you use them together. That’s when they’re the most effective."
To use these techniques together, Moran continues, you can set up a website with a search engine to help decide its navigation choices. "But if it is personalized, you can look at a person’s profile and you can issue different search queries based on who they are. Different people would get different results."
Navigating the Enterprise
The ability to search dynamically has been popular within ecommerce for a while, helping customers find the right items. However, its use within enterprise is growing.
The major difference between dynamic navigation in ecommerce and enterprise deployments, according to Feldman, is that in a retail setting, it is important to connect users with a product in the fewest clicks possible, while in enterprise you have to provide the user with the information needed to find the correct answer as fast as possible. Or, perhaps more simply, in ecommerce an answer is not necessarily the sole answer, though it may well suffice; whereas in the enterprise the correct answer is key.
"You want to make sure people are connected to the information they need," Feldman says. "But most people have problems asking questions so that they get the right answer."
Dynamic navigation, she continues, is really a difference between searching and browsing. "Search conversations don’t let you go back and forth, while browsing engages you in conversation." According to Feldman, there are a number of approaches to dynamic navigation, such as clustering, guided, and machine learning.
Clustering takes all of the documents from different search engines and looks for clusters in the document that are similar dynamically and doesn’t require taxonomy. Guided navigation starts with a set of categories to display results that are pertinent to a query. It helps step the user through the search process. Machine learning describes a process of running documents or information through the search engine to "train it for a particular topic," according to Feldman. As the topic changes, the machine learning system adapts to the new information. It differs from a fixed taxonomy, which, over time, won’t fit a fast-changing area of research.