Integrating Bottom-Up and Top-Down Analysis for Intelligent

The range of hypertext systems continues to expand, from custom-tailored, closed systems to dynamic, distributed, and open systems like the World-Wide Web (WWW). The shift from closed to open systems results in a corresponding decrease in the effectiveness

Integrating Bottom-Up and Top-Down Analysis

For Intelligent Hypertext

James E. Pitkow & Mimi Recker

Graphics, Visualization & Usability Center

College of Computing, Georgia Institute of Technology

Atlanta, GA 30332-0280

E-mail {pitkow,mimi}@cc.gatech.edu

ABSTRACT

The range of hypertext systems continues to expand, fromcustom-tailored, closed systems to dynamic, distributed, andopen systems like the World-Wide Web (WWW). The shiftfrom closed to open systems results in a correspondingdecrease in the effectiveness of metrics and techniques forproviding intelligent hypertext to users. Essentially, thelocus of control shifts away from developers towards users.Stated differently, the central question becomes how cha-otic, loosely constrained environments, like the World-WideWeb, can provide intelligent hypertext. This paper arguesthat viable answers are derivable from both bottom-up andtop-down analyses. Furthermore, intelligent hypertextwithin open, client-server systems may pro t by combiningthese two approaches. Using the WWW as a case study, amethod of analysis for each approach is presented, accom-panied by a discussion of the implications for implementa-tions in open hypertext systems.INTRODUCTION

The proliferation of network and Internet1-based resourceshas dramatically increased the amount of information avail-able to users. Not surprisingly, it has also increased the num-ber of protocols and native browser software necessary toaccess the information. As the amount of information avail-able increases, so does the complexity of discovering,retrieving, and ltering documents and data. Aimed at sim-plifying this task, the World-Wide Web (WWW) [Berner-Lee, et. al. 1992] provides hypermedia access to the widelydistributed and heterogenous collection of existing informa-tion resources via the Hypertext Transfer Protocol (HTTP)and the Hypertext Markup Language (HTML). Currently,WWW enables seamless access to such Internet-basedinformation resources such as Gopher, File Transfer Proto-col (FTP), Wide Area Information System (WAIS), and Net-work News Transfer Protocol (NNTP), to name just a few.Although a precise growth rate is impossible to determine,the dramatic increase in the use of WWW technologiessince inception in 1991 is evidenced by the expansion of

1. For the purposes of this paper,network-based andInternetwill beused interchangeably, though network-based information systems arenot limited to the Internet, e.g., DEC-Net, LAN’s, etc.

HTTP servers (611 in 1992 to over 7,300 in early August of1994) and the number of users accessing each site. In addi-tion, NSFNET backbone byte counts by port number showsWWW as a top ten information service as of August, 1994[Merit NIC 1994].

Nonetheless, technical dif culties remain with WWW, withthe foremost issue possibly being global organization. Whilea complete discussion of the problem is beyond the scope ofthis paper, the most salient rami cation is that without a glo-bal structure, the ability for users to perform global query-ing, (i.e., searches that access all resources in all countriesand domains), and global browsing (i.e., exploration basedupon subject, location, etc.) are not implicitly supported bythe WWW protocol. In short, not only does this make intel-ligent information retrieval very dif cult, but it also does notsupport an interface to the query’s results.

Essentially, WWW provides the underlying structure forglobal hypermedia browsers that provide the interface foraccessing Internet-based information resources. One of the rst global hypermedia graphical user interface (GUI)browsers to be developed was the National Center forSupercomputing Applications’ (NCSA) Mosaic for the XWindow System (X Mosaic). Of cially released in June of1993, initial empirical research estimate2 [Koster 1994] thatX Mosaic currently accounts for roughly 53% of all WWWrelated accesses to HTTP servers3. While accounting forover half of the client access is impressive, NCSA’s XMosaic has also had a large effect on WWW browser designand functionality. This is evident in the new browsersreleased by other developers and companies. In essence, XMosaic has become the current de-facto standard for designand functionality of WWW browsers. Despite this, the inter-face developed does not incorporate intelligent hypertextinterface technology and did not evolve from user-centereddesign methodology.

With the goal of understanding the design of intelligent

2. NCSA’s Mosaic for the Machintosh and Mosaic for Windows al-most evenly account for the another 30% and 14% respectively, result-ing in NCSA’s browsers being used for 83% of current WWW accessesfor the HTTP server being studied.

3. Note that the above usage data support the selection of X Mosaicas the key browser for monitoring user activity.

Pitkow &Recker

GVU Tech Report: GVU-GIT-94-41

Integrating Bottom-Up and Top-Down Analysis For Intelligent Hypertet

Intelligent Hypertext Workshop, Third International Conference on Information and Knowledge Management

The range of hypertext systems continues to expand, from custom-tailored, closed systems to dynamic, distributed, and open systems like the World-Wide Web (WWW). The shift from closed to open systems results in a corresponding decrease in the effectiveness

hypertext for WWW browsers, we took both a bottom-upand top-down approach in our analysis. Speci cally, in thebottom-up approach, we investigated inpidual user hyper-link usage patterns. This was accomplished by modifying XMosaic to log the time and the hyperlink that each user tra-versed. Hype …… 此处隐藏：19182字，全部文档内容请下载后查看。喜欢就下载吧 ……