http:^^www.cs.washington.edu^homes^joebob^web-agent.html
来自「This data set contains WWW-pages collect」· HTML 代码 · 共 152 行
HTML
152 行
Date: Tue, 10 Dec 1996 21:39:56 GMTServer: NCSA/1.4.2Content-type: text/htmlLast-modified: Fri, 17 Mar 1995 21:58:40 GMTContent-length: 5636<html><head><title>A User Interface for Assisted Resource Discovery</title></head><body><h1>A User Interface for Assisted Resource Discovery</h1><p>This document continues recent discussions on the User Interface for anIntelligent Agent that assists in using the World Wide Web.</p><h2>The Problem</h2><p>Finding (and using) resources on a large, loosely-organized network canbe tedious and time-consuming. Intelligent agent software can assist auser, but current user interfaces are either hard to use or do not fullyexploit the agent's capabilities.</p><h2>My Approach</h2><p>I want to construct a testbed forevaluating different user interfaces for assisted resource discovery.The audience consists of casual or "naive" users of information networks(i.e. I don't expect them to be able to program or to know predicate logic).</p><h2>The UI Research Issues</h2><p>What are the User Interface research issues that I want to address?</p><h3>Assisted Search</h3><p>Is providing a software assistant a natural and easy metaphor forgenerating fruitful searches of large, loosely-organized networks?</p><h3>Rapid Convergence</h3><p>What user interface elements of an assisted search contribute toa rapid convergence on a result? How can the user work with the assistantto initially specify, then modify the goals of the search?</p><h3>Specialized Discovery</h3><p>Is it natural to classify <em>types</em> of resource discovery tasks?If so, are there specialized user interface elements that are appropriatefor these classifications?</p><h3>Managing Discovered Resources</h3><p>Once many resources are discovered, how can they be retained andmanaged by the user for future access? What user interface metaphorsfacilitate effective resource management? Such metaphors should integratewith the assistant (e.g. an assistant should keep the resource locationcurrent).<h2>The Domain</h2><p>This work will use a software robot that can understand and navigate theWorld Wide Web. This is not the only domain. It is a very well-knowndomain in which easy, frutiful resource discovery is crucial.In the World Wide Web domain, a most common example of resource discoveryis </em>Search</em>.</p><h2>Initial Work Plan</h2><p>In general this approach applies to all types of resources and to informationnetwork services other than discovery. Ultimately I would like to seethe most effective user interface techniques tightly integrated into a<em>Global Desktop</em> that has built-in software assistants to accomplishmany chores we would normally want to delegate to human assistants.</p><p>As a start I will work on adding an assistant-based interface for resourcediscovery on top of the Mosaic World Wide Web browser. I have consideredtwo scenarios:</p><h2>Scenario 1</h2><h3>A Mosaic Companion Assistant</h3><p>Scenario 1 would provide a companion application to Mosaic that allows a user tosearch for Web resources. Initially this would implement a keyword searchbut it would be built with the idea of adding specialized searches later.The search is carried out by a software agent (so I presume the UIgenerates a <em>goal</em> that the agent tries to satisfy)which returns a collection of Web resources. The user can then<em>tour</em> the results by directing the companion to control Mosaicremotely. While the search is in progress, the user can continue to workwith Mosaic.<p><h3>Advantages</h3><ol><li>A companion application allows the user to remain in control andcontinue working with Mosaic.</li><li>Code that permits remote control of Mosaic has been built locally,or is built into the latest release of Mosaic.</li><li>Future UI experimentation can be plugged in and is not limited byHTML or Mosaic forms.</li></ol><h3>Disadvantages</h3><ol><li>This scenario requires access to code not installed on each client,and not available to a viewer, so use of the project would be limited.</li><li>Since it's not integrated into the viewer, the UI may be seen as a"hack" rather than designing a new UI whose focus is the resource discoverytask.</li></ol><h2>Scenario 2</h2><h3>An Integrated, Assisted Version of MosaicApproach</h2><p>This approach would provided the same search mechanisms as Scenario 1.The UI would be integrated into the Web as another Web page (likeWebcrawler or Yahoo). It would also build a keyword-based query fora software agent, but using Mosaic forms. The results would be returned ina generated HTML document and the standard hypertext navigation wouldbe used to visit the results.</p><h3>Advantages</h3><ol><li>Since this scenario uses the standard Mosaic viewer, it would beavailable to a much larger audience.<li>Form entry and hypertext navigations are widely understood by users.<li>The user does not think of the agent software as an intermediary.</li></ol><h3>Disadvantages</h3><ol><li>This scenario does not provide a fertile UI testbed. Specifically:</li><ul><li>It would be more difficult to investigate UIs for iterative dialogsto rapidly constrain or focus a search.</li><li>It would be more difficult to experiment with metaphors for retainingand organizing collections of known resources.</li><li>It would be harder to automate <em>tours</em> of resources.</li></ul><li>While the agent is working, the user does not have control of theviewer.</li><li>The user does not think of the agent software as an intermediary.</li></ol></body><hr><address><!WA0><a href="http://www.cs.washington.edu/homes/joebob/index.html">Joseph M. Sherman</a> <br>Last modified: Friday, March 17, 1995</address></html>⌨️ 快捷键说明
复制代码Ctrl + C
搜索代码Ctrl + F
全屏模式F11
增大字号Ctrl + =
减小字号Ctrl + -
显示快捷键?