Challenges for Web Information Providers

Continued from page 11 / Link to article's front page

As my list grew, however, a serious usability problem arose. The size of the list, particularly in its HTML form (one very large page) caused problems with some Web browsers (crashing them). Therefore, during the summer of 1994, I modified my translation program to also create a "segmented" hypertext version as shown below.

This segmented version divides the file into various-sized "chunks": sections, subsections, and sub-subsections (which had already been marked in the database file). Using automatically generated tables of contents at one, two, or three levels, users can retrieve just the part of the document they need. This dramatically reduces the amount of download time for a user. Moreover, the separation of the list into files of varying sizes allows the user to select the amount of information they want to encounter. This change was essential for the continued usability of the document itself. The list's size before segmentation was over 150K, requiring large amounts of time to download as well as frustrating the user with information overload.

Improving Content

• Accuracy of sources. In the early days of widespread use of the Net, any information about how to use it or understand it was welcome. Today, the variety of information sources requires me to seek out only those sources that, to my best judgment, are most accurate and usable for my list. The dynamic nature of the Net and the plethora of alternate operating systems, tool versions, and user customization make it impossible to check each source for accuracy. However, I've kept informal track of which sources have proven, in the opinion of others and through my own checking, to be reliable. In some cases, particularly with very new and emerging tools and resources, there may be only a single source of information in existence. Later, as other information is developed, I try to point to resources that are accurate, high-level, and appeal to a variety of audiences (beginners, advanced users, administrators).

• Link freshness. Since Net resources constantly change, keeping links updated in my list is a constant task. Ultimately, I would like to have tools or methods to automatically verify the existence/location of a resource at a given URL. Through an interface to such a a link verification tool, I could verify or correct links in my database automatically. Without such a tool, I rely on user reports as well as periodic sweeps by hand through my list to verify links. I've also developed a sense of which links will be more stable than others (older institutional and organizational links are often more stable as opposed to those developed by individuals or those that are relatively new).

• Reducing redundancy. I seek to link to highest-level, stable, comprehensive information sources for the topics I cover. A positive trend to help me accomplish this is the development of good topic-specific collections of information. For example, I now don't have to offer a wide selection of links to information about developing HTML files. Instead, I can link to a few good collections, including WWW & HTML Developer's JumpStation maintained by Barry Raveendran Greene of the Johns Hopkins University Applied Physics Lab and WebWeavers, a collection of links to assist web weavers, maintained by Chris Beaumont. Some redundancy in my list helps users see alternate views of the same information. However, once an information source appears to have surpassed other similar sources for accuracy, completeness, and usability, I'll tend to list just that resource, particularly when it collects and organizes links to other instances of information that I can eliminate from my list.

• Improving annotations. My database format allows for a short name for a resource and an optional longer description. This longer description gives me the chance to add value to the information by providing a good description of the resource. Later, Oscar Nierstrasz (formerly of the Centre Universitaire d'Informatique (CUI), l'Universite de Geneve) included my list as a source in the CUI Web Catalog, so I am aware that these descriptions play a role in making this catalog more valuable.

• Providing alternate views. I've created other hypertext guides to provide higher-level and alternate views into my list. By developing the Internet Web Text, I've tried to layer the abundance of information so that newer users can encounter it in a variety of ways and in smaller "chunks." By providing narrative, list-oriented and graphical views, I hope to provide users with a variety of alternate ways to learn about the Internet.

Expanding the Context and Activity

With this idea of a need to gather a group of peers, I've created a Web-based forum for sharing information and connecting with other people interested in computer-mediated communication, The Computer-Mediated Communication Studies Center, which includes a resource collection, a directory of people interested in CMC, a list of activities, and a publication, Computer-Mediated Communication Magazine.

By expanding the context in which I develop my list, and by gathering together domain experts interested in the same field, I'm starting to make the important transition from information to knowledge.

Lessons Learned

1. It is possible to develop and maintain a handcrafted index to Internet resources where:
   • The list adds value over what could otherwise be easily obtained (either through another list or index or through space searching tools).
   • The list is continuously updated and improved in terms of selectivity, usability, design, and content.
   • The list is offered in a variety of formats.
2. A single database design using a markup language combined with a translator program is essential to providing multiple formats of a list. It would be far too difficult for me to create and maintain the individual files required for the segmented hypertext version. I would not even attempt to make a single LaTeX version of the file by hand.
3. In order to develop a successful list, the list maintainer should:
   • Keep aware of current developments in Internet resources in the area.
   • Become knowledgeable in the domain area represented by the field of study of the list. The maintainer should also rely on domain experts to help advise on the significance and value of information sources.
   • Be available and accessible for comments from users and domain experts and timely maintenance of the list based on these comments.
   • Provide leadership/vision toward making the list serve the interests of the users by seeking out user opinions and testing the usability of the information frequently.
   • Ask for and acknowledge the assistance and collaboration of others in shaping the information in the list.
   • Actively seek and install new resources, links, or information presentation methods in the list.
   • Provide periodic publicity and announcements about the list to appropriate online discussion forums and indexes.
   • Seek a replacement when he or she is no longer able to develop the information in the list or when absent for an extended period.
4. A resource list exists within a larger context in which its value as information can be used to create or develop knowledge (and hopefully, eventually wisdom). In order to accomplish this, a resource list should be presented and used within a community of people interested in the information, in order to provide the critical review as well as suggestions to improve it.
5. A resource list's ultimate value lies in the judgment and care of the list maintainer. Annotation becomes very important; and, as the information spaces which the list points into become saturated and polluted, the judgment about what to leave out becomes even more important. Eventually, the list maintainer relies on others who create more focused and specialized collections of information to provide excellent destinations as links.
6. It is important to recognize the limitations of Web-based information. In most academic disciplines, important journals, books, and other scholarly materials are still on paper. Net-based information should recognize this, and point the user to appropriate paper-based resources.

Continued on page 13

Copyright © 1994 Sams Publishing. All rights reserved. Printed by Permission.