Technology from Harrod's Creek

Wondir, Wondir, Who Wrote the Book of Pros?

Here's an tough question for master researchers: "Who is the search engine developer who plays in a jazz band?" Here's a clue. This person says the band is "a group composed mostly of geeks from America Online, where I used to work." You have 30 seconds. Start your browser session now. (The answer is at the foot of this column.)

Most online search wizards are likely to find this a difficult problem to solve using any online resource. The truly desperate might look for an expert researcher who can track down our high-profile expert. An amateur might turn to Alta Vista, bang in the word research and slip into a narcoleptic trance pondering the 57 million hits.

One of the most difficult tasks is finding an expert with expertise in a highly specific area. Hitting the precise combination of words to find people who have or display special skill or knowledge is not a simple task.

About.com-usually an excellent starting point for Internet resources on specialized topics-reports that it has 6,923,688 pages in which the word expert appears. Poke around a bit with information broker, consultant, library research, expert witness. What sites pop up?

In 2001, I assembled a short list of 12 human-intermediated search sites. These sites are or were ways to find an expert or allow an expert to answer a question. One year later, I revisited the list and found that three went dark: Expert Central, Frenzi, and Know Post. Atomica, EXP, Expert City, and Web Help shifted their business from providing access to experts to customer service for businesses. Firms in still in the expert business are Abuzz, All Experts, Ask Me, Experts.com, and Keen.com. Most of these companies continue their long day's journey into plight.

Now enters Mothra; that is, Google. The expert-for-hire niche is now more crowded. Google is viewed as one of the top two or three ways to locate information on the Web. Google wants to do for experts what it has done for search.

Getting an answer to a question, however, need not cost much if any money.

The Library of Congress offers what is a global, collaborative "reference service called QuestionPoint." Researchers accept questions via electronic mail and provide answers. The Library of Congress worked with the U.S. charity OCLC, Inc. to develop the service, which according to a Library of Congress spokesperson is "used quite a bit by researchers all over the world." Answers are provided by a Global Reference Network, which is a group of libraries and institutions worldwide that are committed to digital reference.

The quite useful Research Buzz online newsletter called my attention to the National Library of Scotland's new research service. One completes an online form at www.nls.uk/info/readingrooms/askalibrarian.html.

Librarians of the NLS will respond directly with an answer or share the question with another library should more research horsepower be necessary. NLS alerts those sending questions that a response might take some time, but the Web site offers a tracking feature. A click shows the person submitting the query a status report.

Another resource is Experts.com. There the answers cost money. Experts.com is essentially a directory and brokering service, and it is one of the best of the expert finders and for-fee help sites available at this time.

New search services and tools from Teoma, Vivissimo, and Wisenut offer useful ways to perform queries, but finding experts requires the same slogging through large numbers of hits. Clustering, associative functions, and other powerful features are helpful but not comprehensive solutions.

Dr. Matthew (Matt) Koll, one of the pioneers of online search and retrieval, has set up a not-for-profit foundation to make finding experts easier. After starting Personal Library Software and selling it to America Online, Dr. Koll is embracing the open source approach to database building.

His approach is an innovative blend of data, software, and service. Dr. Koll says, "Finding experts is difficult. It occurred to me that creating a collaborative community was a new approach to building a database. The software I needed was not available in open source, so I decided that my team and I would create a suite of tools to provide the functions we needed to build this database. In order to make sure that the base software is enriched, we are offering this as open source software. We want to create a model that supports filling a void in databases and involves a large number of people in making the service vital and fresh. My thought is that the value of the service is greater than the sum of its parts."

Dr. Koll, one of Gerald Salton's prize students at Cornell University, operates the new service as Wondir.

He says, "Google, FAST, AltaVista and other services are doing a very good job at serving common and popular needs, even specialized needs a lot of the time. But the better they get at figuring out what most of the users want they ask any particular question, the harder it becomes for all users to find the answer they need -- especially if it happens to reside in a less-popular or less-visible sources. Sources that might provide the best information for some users for some questions just might not have the popularity to make it to the top of the hit list on the major services."

"As the search and information business grows, one-site-fits-all becomes less and less viable. Community services, target services, people who want to make their information available to their constituencies find it harder and harder to be found on the mass-market services. By putting the focus on people answering questions, we'll make sure that a broader range of voices is present on a search results page."

Unlike the services on offer from directories, Wondir is designed to accommodate anonymity if required by the questioner or the expert. Dr. Koll says, "What is indexed directly at Wondir are the profiles of people. One of the more compelling benefits of a system of Wondir's design is that experts need not reveal their personal information. Their electronic persona will be judged on the quality of the answers it provides-judged by the users. Some experts can choose to reveal their identities and/or provide their credentials. Many experts will be credentialed by professional associations, so users will have a higher degree of confidence. Some experts will not be individuals at all, but rather question-answering services, like the AskA services at www.vrd.org. And there is also a role for anonymous experts. I can imagine, for example, that a political refugee providing answers and guidance back to people suffering under a repressive regime at home, might want to remain anonymous. Likewise, a doctor providing information about reproductive choices might not want anti-abortion zealots to find out who he is."

Most information industry executives have come to the conclusion that the economics of database construction and online search and retrieval are reasonable challenging. Money is generated from commercial licenses and selling traffic by auctioning words. The more popular a word, the more clicks it generates. Selling hits is slowly and steadily eroding the objectivity of search results. Dr. Koll says, "The commercial services are under a lot of pressure to monetize the results page. Between advertisements, paid results and inducements to lead users to other money producing services, the space available for unbiased results is pretty limited. Wondir is a non-commercial entity. We can keep more of the results page open to the wide world of good sources. And when people see a resource on a Wondir page, they'll know that its placement was not influenced by money changing hands."

One way to locate experts is to turn to directories of members of such organizations as the American Institute of Aeronautrics and Astronautics yield superior results. However, access to the member profiles often requires joining the organization or a trip to the charity's library. Most know that a print directory's data are stale when the book goes on sale. Some directory publishers are quick to sell advertisements and somewhat less nimble in producing updated versions of these compendia. Vanity directories are more amusing than useful, particularly those purporting to identify who's who in a particular region. For-fee online directories fall somewhere in the grey zone of unknown authority for the "facts" they contain.

Dr. Koll, developer notes, "Anyone can say he or she is an expert. Determining if that person has the expertise you require is a very difficult and time-consuming process. There's a need for a more authoritative expert-finding service."

He adds, "In most places in the world, if you ask someone a question that is within their expertise, you get a good answer, cheerfully and helpfully delivered, and both people go off feeling satisfied. That's true on a construction site, in a library, on a farm, in business, at a big city intersection, in rich and poor communities."

Wondir will include top notch specialized databases from the invisible Web; that is, the Web pages not indexed by commercial spiders. One feature of the Wondir software allows an expert to submit a factual profile without compromising confidentiality. Dr. Koll observes, "We want to tap the power of a community of professionals and their feedback to help us select the most appropriate experts and other sources. The Wondir effort is about community, about grass-roots movements. We're providing a place where community organizations can get the word out. We anticipate a community of volunteers helping to develop the technology. We view the technology, indeed the whole service, as community resources. This is something that we believe belongs in the commons, for everyone to help build and to use."

One of the unique aspects of the Wondir project is the use of search technology so that the user and the system combine to "push" questions to experts. Mr. Koll notes, "Many of the expert finding systems require the experts come browse all the questions. Some pushing each question to everyone in a large category. Wondir-and this is new and pretty elegant-uses software to control and direct the flow of questions to experts. And we are adding geographic filtering and weighting to our algorithms. The result is more flexibility and a faster turnaround on questions."

The most important innovation in the Wondir expert database is the emphasis placed on intelligent agents. The Wondir system is designed so that it learns over time from the questions and answer. The Wondir system suggests experts to the user. As the users interact with the Wondir system, agents will learn from user actions and become increasingly precise in their display of experts and the system-generated recommendations."

The business model for Wondir is an about-face from the commercial business models of Personal Library Software and America Online. Mr. Koll observes, "Wondir does not impose any economic view on people. We'll encourage free information. And the results will be totally free from commercial influence. That said -- if some answer providers have value that is worth paying for that's fine too. We'll always point out if a source is charging for its information. But we'll let the marketplace decide if people are willing to pay for it. If a meteorologist or a mathematician is solving a complex business problem with far reaching financial implications, then those experts should expect to be paid. But if they're explaining a concept, guiding or encouraging a student, they generally don't want compensation. Wondir will not dictate how those participating in our project must deal with questions and those who ask them."

Wondir is designed as a global service. The objective is local content that covers the world. Dr. Koll sees Wondir as including experts representing virtually all disciplines and subject areas. The experts themselves will, he says, "span the world, and its languages."

Dr. Koll emphasizes, "The vision is that anyone, anywhere in the world, at any time can simply ask a question -- and that someone will be online to give him the answer he needs. Organizations that have a community outreach mission will adopt Wondir as their service. Beyond that, its hard to predict. If we're successful, communities-of-use will spring up and who knows how they'll use it, what directions they'll take it in."

The answer to the question at the head of this column is Matthew Koll. He insists he is an average musician. He may not be Miles Davis, but he is a Wondir.

Stephen E. Arnold

Postal Box 320

Harrod's Creek, Kentucky 40027

Voice: 502 228 1966

Electronic mail: sa@vnu.co.uk

Web site: www.arnoldit.com

Mr. Arnold is an independent consultant working from Harrod's Creek, Kentucky, USA with two boxer dogs and new Xeon servers. His new book is Knowledge Management Sense and Non-Sense: How to Make Data Payoff. It will be published by Infonortics, Ltd., Tetbury, Glou. In early 2003. Mr. Arnold specializes in competitive intelligence and technology analysis.