Currently, there is a lot of innovation going on in search. Not only in technology, but particularly in what customers are doing with search. Search is no longer just about the familiar search button, but has become a central platform providing situation-specific information and user interfaces to customers and employees. This is echoed in the marketplace, where search vendors no longer are only speaking of features under the hood such as relevance, operability and security. The new mantra is delivering "the answer" in a time-critical manner within the context of the individual. The technology supporting this claim is the search platform’s provision of user interfaces with information mash-ups from disparate systems tailored to the individual’s wants and needs. Put differently, the search platform has become a broker between user context and the available data and applications.
The emerging technology opens up a raft of new search adoption opportunities, of which we have just seen the beginning. Customers—with their willingness to experiment with emerging technology and their in-depth knowledge of their users—have an important role in uncovering these opportunities. Some adoption trends of the emerging technology seem to crystallize. These are:
- Dynamic consumer portals where search technology is the platform connecting people and information and driving user interface mash-ups;
- Corporate intelligence solutions; and
- Functional and personalized knowledge management solutions.
Common to all these adoption areas is the search platform’s role in providing the users with precise situation-specific information and functionality. This role requires the search platform to understand the user’s context in order to give precise answers—a task that requires contextual search. Contextual search is an emerging technology that is still in its infancy and requires further refinement.
In this article we will focus on the search platform’s role as a match-making broker between user context and the available data and applications, and in particular on how contextual search will need to capture front-end interaction and end-user context in order for the search platform to deliver on its promise of delivering "the answer."
The Case for Capturing Front-end User Context
Historically, as customers have demanded better and better search capabilities, vendors have typically responded with more and more back-end features. When users wanted broader hits on their two-word queries, we responded by adding synonyms. When they demanded better accuracy on misspelled queries, we responded with a wide variety of bells and whistles to correct the query or suggest new ones. When more and more repositories got added and the data volumes grew, we responded with faceted search and drill-down capabilities for quick navigation. For each new demand, a new back-end functionality was added to alleviate the problem, always moving search technology a little bit forward.
Contextual search now accelerates the pace of change. Its ability to capture user context has the potential of creating a new quantum leap in the industry. This new product direction is already evident in many of today’s vendor solutions. Vendors, including IntelliSearch, have been talking about relevance and contextual search for the past year or so.
Take MOSS 2007 as an example. It rates documents closer to where you are standing in the site hierarchy when searching. This is a good attempt at trying to capture the user context in order to improve search precision. But what happens when you move outside of the SharePoint search box? All context is lost. The same goes for Outlook. The system knows you’re searching emails and will limit it to emails, sorted by date, assuming more recent emails are more relevant. As for Google, it does not capture the context of what the individual user is trying to achieve at all, but uses other users’ experiences and guesses that you are doing the same as them. The first user to search for an imaginary new car named "Black" will not find anything until people start talking about it and linking to it, even though the site might be indexed. If you knew that the user previously searched for this car and also knew that she had visited the sites of Ford and Chrysler, you could add the context "car" with the word "black" and get a much better result. If you also attached the context "new" to the search because the user was chatting with a friend about buying a new car, you would get even closer.
The Desktop—Essential in Capturing Context
Only the desktop, and not your standalone application, will know what you do at any given time, as it is where you do all your work. It can capture all your input and monitor the chain of events as you work on a problem. The desktop is simply the best available avenue to capture user context and to improve the user’s search experience. Further back-end improvements are by no means unimportant, but without capturing the front-end interaction, the back-end can only achieve so much.
For instance, by logging user behavior linked to time of day, the search platform can identify your information needs pattern and context. How much time do you spend on reading documents from Bill compared to a document from John? Did you just read four documents about a new customer before you did a search? Are your email conversations linked more to people in your department or with customers outside the company? Or are you currently working in your customer service application?
One may argue that using all these contexts is problematic since they potentially point in very diverse directions. However, in real life most people only work at one problem at a time, so within a defined time-frame all your context fragments would point in the same directions. Even if one of them might be way off, all the others would pull you back in, still yielding a much better result than you would without the context information.
Search Platform Imperatives
The whole purpose of including front-end context is to improve search precision. By adding front-end context into the search processing you can dramatically improve result precision on both a corporate and functional-specific level, thus enabling precise and fine-tuned information access supporting a multitude of departmental needs such as customer service and engineering departments.