Q: Let’s just jump right into the basics: What makes enterprise search so difficult? We keep hearing this question: Why can’t it be as simple as Google?
Peter Morville: Google is able to draw on the page rank algorithm, which looks at the links between websites as implicit votes of confidence. The most popular sites from a linking perspective float to the top.
At the enterprise level, you’re dealing with a smaller set of users. And, they don’t get as excited and enthusiastic about their enterprise’s content, so they’re not as willing or able to do all the cross-linking. Hence, the tendency is to fall back on the older model of relying on the words in the text and supplementing that with metadata.
There are additional challenges unique to enterprise search.There are so many different file types. Plus, search has to work across multiple departments and business units—and it’s not at all easy to reconcile the different goals, priorities, and perspectives that come with multidisciplinary efforts. You also have cultural and international challenges.
But, there are a number of things you can do to improve the situation: tune the search engine’s relevance ranking algorithms to the content; carefully design the results interfaces; take a close look at the content and make sure to remove the content that’s redundant, outdated, or trivial (also known as the “ROT”). If you can make the information space smaller, then the search is going to be a lot more successful.
Q: You mentioned metadata. Can you speak to automated versus manual tagging?
PM: On one hand, it’s great to be able to automatically extract certain types of metadata.Things such as document type, author, and creation date are often easy to capture. But machines don’t do so well with topical or subject oriented
classification. They don’t understand what we librarians call “aboutness.” So, people should be involved in tagging some metadata fields, at least for the most important content.
Q: There’s a lot of talk these days about guided navigation, can you help the readers understand what’s involved?
PM: The first successes with guided navigation occurred in the e-commerce market.Take, for example,Tower Records. You might enter a couple of keywords for a song whose name you know.You get back a massive results list, but on the left side of your screen, you see a narrow panel that provides you with metadata fields and values that provide an opportunity to further refine or narrow your query. So, you might limit this search to albums from the 1970s, or by genre, artist, or price. You can perform this narrowing in an iterative, interactive fashion until you find what you need. This approach is just starting to succeed in enterprise settings where there’s more work to be done to define those metadata fields and tag the documents, but there’s a real payoff from a findability perspective.
Q: What about any problems involved with a guided-type navigation solution?
PM: The biggest challenge in the enterprise setting is tagging the documents. So, there’s a good amount of heavyduty information architecture work to be done in defining the metadata fields and developing the controlled vocabularies or the values for each field. It is difficult but not necessarily all that time consuming. But how do you tag all this stuff? That takes us back to the earlier point, whereby if you’re able to do a decent job with automatic metadata extraction, you might be able to tackle most of your content. However, the real high-value topical tagging is almost certainly going to have to be done by people.
This is where you really have to get into defining different tiers of content, and perhaps manually tag only a small percent of the most valuable enterprise-level content.That way, you’ll provide better access to that core content while the remaining content will still be accessible via traditional search.
Q: Let’s talk a little about your book and where you think search is headed.
PM: Ambient Findability is very much focused on the future. At the crossroads of ubiquitous computing and the internet, I see us heading toward a world where we can find anyone or anything from anywhere at anytime. Technologies such as GPS, RFID, and cellular triangulation are making it possible to tag and track high-value objects and each other. As we create an “internet of things” populated by location-aware devices (not to mention sensors), the concepts of retrieval and wayfinding are converging in some interesting ways. As the book’s subtitle suggests, what we find changes who we become…
As a librarian, I’ve long been aware that the iterative and interactive nature of search involves a lot of learning. In fact, I would argue that in this age of knowledge work, search is among the most important ways that we learn.We enter a few key words, find a document, read it, and discover that we have been searching for the wrong thing. So, we try a different search term, and we find an expert in the field. Perhaps we hire them as a consultant, and they teach us something else and send us in new directions. Ultimately the search shapes our goals and what we believe.
Q: How prepared are enterprises to fully embrace the potential of search?
PM: It depends. I see a huge discrepancy between the few organizations that really get it and are way ahead, and the other organizations that are back in the 1990s with respect to their enterprise search systems. So, in the near term, we have an awful lot of work to do to simply make our existing systems work better—sometimes organizations take two steps forward and three steps back. It’s not uncommon for companies that integrate new content management systems and search engines to actually degrade the user experience in order to accommodate the new software.
In the longer term, I’m intrigued by the convergence of physical and digital systems and experiences. Organizations will have the opportunity to become more efficient by tracking a diverse inventory of people, information, and objects. Hospitals are already using a wireless product from Cisco that lets them track the location of their wheelchairs. And they’re showing a substantial return on investment, because they’re saving tremendous amounts of staff time—instead of searching around for
that wheelchair that got stuck in a closet somewhere, they’re able to locate it immediately.
There are opportunities within many enterprises to tag and track high-value objects and really improve efficiency and customer service at the same time. As the digital and physical world converge, we’ll identify products and solutions we’ve never considered before. It’s an exciting time to be in this business. As science fiction writer Bruce Sterling noted, “the future isn’t just unwritten—it’s unsearched.”