Enhancing personalized results is a large near-term goal for Web search, according to major search-engine companies. Other prominent objectives include improved multimedia searching, localization, question-and-answer technology, visual results presentation, and modified Web search on devices other than PCs, including cell phones.

Take a peek at future plans for search engines such as Ask Jeeves and Google, and you'll see a vision for Web search's future that's more sophisticated, individual, and portable.

Which changes will be key? Personalization, localization, and improving question-and-answer technology are equally important to improving daily search experiences, says Amanda Spink, a professor at the University of Pittsburgh's School of Information Sciences and a longtime researcher of Web search behavior. "Each technique relates to a different aspect of peoples' information behaviors," she says.

There's plenty of room for innovation. Despite a slew of features improvements and results presentation experiments, individual Web search behavior hasn't changed greatly over time, according to research in Spink's new book Web Search: Public Searching of the Web (Springer, 2004), which looks at search engine use from 1997 to the present.

Hurdles remain to making personalization and localization of Web content common, both on the user acceptance side and the software development side. But if current R&D projects in these areas succeed, search engines will be better equipped to give searchers exactly what they want on the first search try.

"Our challenge is to read a user's mind," says Daniel Read, vice president of product management for Ask Jeeves. It's an intriguing challenge, given that most Web searches today still contain just two to three words.

Personalization holds both great promise and possible pitfalls for search engines. Wouldn't it be great if a search engine, by getting to know you over time, could remember that you frequently search culinary topics, for example, and could filter your results accordingly? A search engine in this example would learn that your queries regarding fish terms are most likely related to cooking, as opposed to the sport of fishing.

On the flip side, some users have privacy concerns about a search engine knowing too much. Still, the big players, including Google and Ask Jeeves, are trying hard to make personalization work.

To date, search engines have been reluctant to make users supply more information about themselves; this presents a large technology obstacle, Spink says.

"Research suggests that you need to focus on a combination of implicit and explicit personalization," Spink says. "One of the biggest obstacles is that computer software is not very good at really understanding peoples' information needs. Using just implicit techniques is not effective, but explicit techniques need to be much more developed."

Google is trying to give users better customized results based on voluntarily supplied information. Google's Web Alerts let you sign up to receive email alerts when new information tied to your search terms posts on the Web. The company's Personalized Web search (http://labs.google.com/personalized), still in beta testing, lets you create a search profile to filter results. And, Site-Flavored Google Search, also in beta testing, lets Webmasters for sites that use Google as the prime search engine create the profile for their individual sites. The data is then used to tailor Google searches done on those sites.

Ask Jeeves hopes that its service, perhaps best known for its question-and-answer technology, will also become known for personalization. In September, the company began beta testing its MyJeeves service, which lets you save and organize search results and related content into your MyJeeves folders. You can also stash personal notes in those folders. If you register, you can save an unlimited amount of information, accessible from any PC; if you don't register, you're limited to 1,000 documents.

The next step will be for Jeeves to connect MyJeeves with its existing Smart Search features, which return structured data for common questions, such as the time in a particular city. Type in "NYC" at Ask Jeeves, and you get a Smart Search box that includes links to maps, jobs, weather, local time, and the city's chamber of commerce.

"We want to combine the structured and unstructured data," Read says. "To me that kind of completes a circle in search."

But to really succeed with searchers, Read says Ask Jeeves has to do much more than offer a good mix of structured and unstructured data, and personalization is key. "We need to understand the intent behind 'NYC,'" he says.

Ask Jeeves has a team of people in its User Needs and Behavioral Research program studying these questions of intent. Their data analysis includes ethnographic research, qualitative research, and measures such as users' pick rates among search results.

What's the hardest part of the data analysis? "All of it," Read says, adding that understanding intent is a tough area.

He gives the example of someone who does three months of online research before buying an Apple iPod music player. If that person does an iPod-related search, how can Ask Jeeves know where he is in that research process? In the future, MyJeeves might be able to make intelligent guesses based on the user's search history.

Read calls this the "triangulation of user intent." The triangle includes your past search information, other people's previous searches, and your current query. The eventual goal is to craft Ask Jeeves searches to reflect your interests and history. So, for example, someone who frequently searches about travel would get different results than someone who frequently searches about history when typing a query like "Chicago" into Ask Jeeves.

"As we drive toward it, we have to give the user control," Read says, in terms of how much information to supply. "We'll have to ensure that consumers are happy with the search transaction," he says, the same way people gladly trade information for store discounts or rewards in loyalty card programs.

To tailor results to your locale, a minimally invasive alternative to having you fill out a registration form might be for Ask Jeeves to offer to remember your zip code, Read says.

The company hopes that localization of this kind would narrow a wealth of searches for users. Today when search engine developers talk about localization, much focus is on yellow-pages style information, such as store and restaurant listings, but that's just a start.

Another example of a search that's tough today but should soon get easier is the hunt for local professionals of a certain kind, say certified public accountants or master electricians. Type "electrician" plus your city name into most search engines, and you're not likely to receive what you want: a tidy list of local electricians. Google is trying to achieve this through its Google Local service, in beta testing at http://local.google.com. It works particularly well for searching for local restaurants.

In a similar effort, Ask Jeeves started beta testing its Ask Jeeves Local (http://web.ask.com/local) in September. In addition to local businesses and information, Ask Jeeves Local delivers related user ratings and data from Citysearch.

Some search engines are thinking truly locally—as in your PC's hard drive. Search engines could help more people get a grip on data hidden not only in the Web but also in their email and document folders.

Lycos' HotBot was ahead of its peers in this area. Launched in February of this year, the company's free, downloadable HotBot Desktop toolbar was first with an integrated desktop search tool for the Internet, selected local PC documents (such as MS Office application files and .pdf files), and email folders. HotBot Desktop (www.hotbot.com/tools/desktop) also bundles in a pop-up blocker and an RSS news reader. Other companies, including Google with its Desktop Search (http://desktop.google.com), have since weighed in with integrated search options. Microsoft, which recently launched the beta version of its MSN Search (www.msnsearch.com), reportedly plans to release its desktop search tool before the end of the year.

Online retailer Amazon's search engine subsidiary, A9 (www.a9.com), is taking a slightly different approach. The site, which uses Google's core search technology, lets registered Amazon users save searches for later viewing and further keyword queries. It also lets users save bookmarked lists of sites and a "diary" of notes related to particular sites. Users can also perform searches that query multiple sources at once, including the Web, personal notes, selected databases and books from Amazon, and an online dictionary.

Another answer to "What's next in search?" moves beyond the PC. Specifically, as the popularity of Web-enabled cell phones grows over the next few years, search engine developers must adapt their products to fit this platform's needs and limitations, including small displays. Google is testing its Froogle shopping service on cell phones, and Ask Jeeves thinks cell phones will be key to the future of Web search.

One example: Ask Jeeves envisions a day when it will become common for you to use your cell phone to search for a local restaurant and then click a link to make a call to that restaurant.

Once voice recognition technology and local search information improve a bit, Ask Jeeves also sees you asking the built-in GPS unit in your car to find the nearest pizza place and give you directions.

"In the future, search becomes an operating system for peoples' lives," Read says.

Today, you can perform Web searches through devices other than your PC—including searches using a Blackberry PDA or the Comcast cable system, for example—but the searches don't work the same, Read says. You can get localized Web search information on a cell phone, he says, but no search engine has mastered it. Hardware manufacturers, service providers, and search engines have to work together to display the information better, he says. For developers, this goal means thinking about how to organize and display data on multiple devices, not just the one at hand.

While Read sees the restaurant example becoming possible within a year, he says more options are about three years out. Ask Jeeves wants to make these away-from-the-PC searches widely useful: If you search for a local pizza place and it works well, but you search for a local plumber and it doesn't work well, that's not acceptable, Read says.

Also consider new kinds of data indexed through Web searches. "Where we think search is going is aggregating other data sources for you," Read says. As data improves and can be transferred more easily (using technologies such as XML) and homogenized, developers at Ask Jeeves will be able to add data such as TiVo or Web log information, he says.

Better visual presentation of search results, including the clustering of related categories of information, will also have to play a role in the near-term future of search, Spink says. She's just finished a research project for Web search engine Vivisimo that tackled these issues.

"Overall, visualization techniques have not been successful and are very preliminary," Spink says. "They are currently far too complex for most people to understand."

What would she like to see happen next in this arena in order to make searching more effective?

"Beyond visualization, we need to reexamine our assumptions about the presentation of search results," she says. "For example, why can't the results be grouped according to different attributes such as date, relevance level, or other criteria? Web search engines are offering very limited options in users' ability to resort the results."

Searchers themselves might also need to be held to a higher standard, Spink says. In other words, people might want to learn advanced search techniques.

"The Web search engines have made some changes, but one major problem is that people are not investing time in learning to use search engines more effectively, and the search engines have done little to help them," Spink says.

"To use electronic information effectively, people need to understand what their information behaviors are and how to improve them and develop more effective complex behaviors. The focus by the Web search engines has been towards keeping everything as simple as possible and trying not to make cognitive demands on the users," she says. "I believe that, in the long run, this will not prove fruitful in helping people do better searching."

Despite the runaway success of Google, market research shows that users want multiple search engines. A May 2004 report from market research firm Vividence, examining how people interact with search engines (based on the search behavior and opinions of 2,000 people), found that Google's primary strengths are its brand marketing and its presentation of results, as opposed to the relevancy of its results. Although 76 percent of searchers reported having a "primary" search engine, and Google and Yahoo lead the pack for user loyalty and repeat usage, up to 47 percent of searchers use an alternative when they don't like the results. Up to 20 percent of searchers in this study reported regularly using specific search engines for varying types of searches. Bottom line? Much opportunity remains for creative developers and designers to make a mark on Web search.