by Christoph G. Thomas
The principal techniques to retrieve information are searching and browsing. In the World Wide Web, these techniques are supported by search engines and bookmarks. Both are invaluable tools when searching and browsing the Web. Moreover, bookmarks allow users to personalize the Web, ie, they create their own personal information space as their "image" of the World Wide Web.
As the World Wide Web is a living "creature" that evolves and grows continually, users have to take care that their personal information spaces can be kept manageable and up-to-date. This is not an easy job because users have to find the right balance between adding new bookmarks, deleting unused or seldom used bookmarks, and updating existing bookmarks. For example, users often struggle with old or invalid bookmarks and have no idea whether their bookmarks still represent a valid image of the World Wide Web.
With BASAR (Building Agents Supporting Adaptive Retrieval), we have developed a prototype that assists users when managing their personal information spaces by:
Software agents called web assistants provide support for these tasks based on a model of the usage (preferences, interests, and tasks) using both explicit (asking the user) and implicit (observing the user) modeling techniques. BASAR embodies the following two features:
• Active views: BASAR provides users with a method for creating personal information spaces with a semantic meaning that is independent of the World Wide Web-viewer (such as MOSAIC or NETSCAPE). An active view is defined as a set of bookmarks belonging to the same header in combination with a set of web assistants attached to it:
The prototype BASAR has been implemented in Smalltalk as a multi-agent system. It is available upon request.
Please contact:
Christoph G. Thomas - GMD
Tel: +49 2241 14 2640
E-mail: christoph.thomas@gmd.de
return to the contents page