August 1995 (Vol. 10, No. 4) pp. 42–43
0885-9000/95/$31.00 © 1995 IEEE
Published by the IEEE Computer Society
Published by the IEEE Computer Society
Guest Editor's Introduction: Harnessing the World Wide Web
PDFs Require Adobe Acrobat
Unless you have been in a coma for the last 12 months, you could not helpbut be aware of the explosion of interest in and use of the Internet. Startingwith this issue, IEEE Expert will run a special series focused on IntelligentInternet Services. This will not be yet another overview of the World WideWeb. Rather, we will present an exciting mix of technology and applicationarticles on how intelligent system technology is harnessing the vast, unstructuredinformation space afforded by the WWW.
This series will provide highly visible, realworld examples of new andemerging intelligent systems techniques. These articles will use several casestudies to illustrate what can be done today and will also highlight futureopportunities. We'll provide ample pointers to Internetaccessible informationsources.
Here is a short overview of the articles you'll find in this issue:
- Folks attending the most recent IEEE Conference on Artificial IntelligenceApplications discuss the results of the special workshop on AI in the WWW(pp. 5055). The report by Carol Brown, Les Gasser, Dan O'Leary, and AlanSangster provides an overview of "supply and demand" agents on theWWW, as well as a discussion about AI information on the WWW. Inaddition, they provide an excellent set of references to other indepth papersin these areas as pointers.
- The topic of information agents providing intelligent Internet services receivesspecial attention in this series. The "Executive Insight" department (pp. 69)by Sara Hedberg describing the emerging capability of information agents isa interesting, accessible look at the present and future of agent technology.
- In their article on intelligent information agents that are currently in use onthe Internet, Oren Etzioni and Daniel S. Weld of the University ofWashington describe the architecture and development approach used toprovide scalable services to a wide range of users (pp. 4449). Currently,they are using feedback from users to refine and improve the intelligentagent services.
In upcoming issues, the Intelligent Internet Services series will continue withthese offerings:
- In their article about satellite imagery dissemination via software agents,Christopher Toomey and William Mark of the Lockheed Martin AI Centerdescribes intelligent information services geared to providing usersinformation, rather than data. The domain of satellite imagery provides aninteresting case study because of the vast amounts of imagery and otherdata coming constantly from a variety of remote sensors, and because ofthe increasing importance of the information that can be derived. This articleemphasizes a userbased definition of intelligent service, paying primaryattention to design and implementation methods that focus on flexibility andutility of information for the user.
- The domain of information access and creation is critical to most computerapplications for the US Department of Defense, government, and industry. Inhis article about effective text retrieval, W. Bruce Croft of the University ofMassachusetts, Amherst, provides realworld examples of documentrepresentation, query processing, and textretrieval techniques used andevaluated in the domain of document detection. Although these techniquesare not knowledgeintensive, the results are excellent and the performance isperceived to be intelligent because the system can be trained to avoid majormistakes.
- Another article by David H. Jones of IndustryNet and Navin Chandra ofCarnegie Mellon University describes a model for electronic commerce thatfocuses on the economic value of effective information access andmanagement. Again, the techniques used are perceived to be intelligentbecause of the value they provide the user.
- Some folks want to talk with their agents and have a dialog much like onewould have with a colleague over the telephone. The emerging ability forcomputer systems to hear and to read, and to understand(!) what they hearand read will greatly influence the definition of intelligent services in the nextdecade. In his article on conversationbased systems, Victor W. Zue of theComputer Science Lab at MIT describes the architecture and function of asystem that provides information access over the Internet. These excitingresults portend a change from a windowsbased, desktopcomputer computingmodel, to one in which people can affordably access information over thetelephone from anywhere in the world using speech and language as theprimary access method. When this happens, the network does become thecomputer.
- To provide insight into policy and technology, the MIT AI Lab will alsodiscuss how they worked with the Clinton administration to bring electronicmail and other services to the White House along with more advancedservices currently under development. This article will provide examples ofhow the use of information technology can allow federal policy makers toinform a broader range of individuals by providing information, guidance,and explanation about regulation and government activity.
We hope this special series will encourage ongoing discussions among ourreaders about the potential of intelligent Internet services. We hope as well thatthe more adventurous among you will be moved to undertake implementationefforts of your own. The recent attention paid to the interoperable architecturesthat underlie snapin modules suggests again that meaningful accomplishments inthis arena are just beginning. Many exciting developments lie just over thehorizon.
Perhaps most surprising has been the rapid pace of progress in this area ofthe computing world in recent months. We encourage you to document yourown accomplishments with these services and to share them with our readers.We'd love to hear what people are doing with these services, and whatlessons they are learning along the way—especially from the aspect of theintelligent systems behavior involved. Together we can make IEEE Expert aforum for the exchange of such war stories.
Jay Allen Sears is a program manager for human language systems (HLS) andhumancomputer interaction (HCI) at the US Advanced Research ProjectsAgency, and also is a senior research manager responsible for developingresearch and technology transfer strategies for a wide range of informationtechnology programs at the Corporation for National Research Initiatives. HisHLS program develops methods for computers to hear and read, and tounderstand what they hear and read. The HLS program involves informationcreation and access for applications such as crisis management, health care,and intelligence analysis. He received his BS in naval engineering from the USNaval Academy in 1966, an MS in computer science from the Naval PostGraduate School in 1972, an MS in business administration from GeorgeWashington University in 1979, and his PhD in computer information systemsfrom the University of Arizona in 1982. He is a member of the IEEE ComputerSociety and the ACM. He can be reached at ARPA, Arlington, VA;asearsarpa.mil.