Delivering internal information is a challenge in many organizations. Information, particularly in the form of documents, is often spread throughout an organization in a haphazard manner. People have difficulty finding the information they need, often because an assortment of inconsistent tools has been developed to access documents. As text is managed electronically and delivered online, these problems can become even more evident.
To address information access and delivery problems at SAS Institute Inc., a creative internal application of the World Wide Web was developed. This application provides a consistent interface and simple access methods to the Institute's pool of diverse documentation. The resulting application is called the SAS Wide Web (SWW), a local, completely self-contained version of the World Wide Web. With the Mosaic viewer as a common front end, and with open access to employees, the SWW has grown in popularity and usefulness at the Institute.
This paper presents a case study of the SAS Wide Web. The first sections describe the history of the SWW project and the current internal SWW set up. The next sections examine the current uses of the SWW, concentrating on the tools used for creating web documents, the challenges the document developers have faced, and the successes and difficulties that have thus far been encountered in the SWW's several months existence. The paper concludes with a discussion of the future directions for internal information management with the SWW and other possible uses of the web. Overall, the paper demonstrates that, particularly as an information integration tool, a localized web application can be an asset by significantly improving internal methods of information distribution, access, and presentation.
Like many organizations, SAS Institute has found it increasingly difficult to deliver internal information. The problem developed over time; more and more documentation was created and stored in a variety of formats requiring a diverse tool set for access. People needed to find these resources quickly and easily to get their jobs done effectively, but the plethora of tools often got in the way. This problem is remarkably similar to the global problem of information access across the Internet. Institute developers observed how the World Wide Web (WWW) answered the global problem of information delivery on the Internet and built an application that took a similar approach for the local problem. This application, called the SAS Wide Web (SWW), has been met with enthusiasm at the Institute, as it has successfully integrated many sources of online documentation in a single access method. With the SWW, employees have access to a widening pool of online information that is presented in an easy-to-use format, via NCSA's Mosaic. This paper describes the evolution of the SWW, from its origin to its current state, and shows how it has become a valuable tool in a short period of time. As the SAS Wide Web exemplifies, WWW technology can be applied effectively on a very small scale, and it benefits the local organization in much the same way as the WWW improves upon the Internet.
The Internet provides a wealth of information useful to professionals and researchers in almost every field. Finding that information, however, has always been a challenge. Over the years many different tools have been created that help people locate and use the information easily. Due to the abundance of tools, a wide variety of interfaces have to be used to retrieve information. The World Wide Web project started by adding yet another information format and protocol, the HyperText Markup Language (HTML) and the HyperText Transfer Protocol (HTTP) server. This effort was different, however, as the developers realized that the value of other information sources would be multiplied if their disparate protocols could be accessed from a common interface. These developers included the ability to create links to other information services from HTML and built a client library that would handle the multiple protocols. Thus, the Web clients have the ability to interface to many of the existing services on the Internet, giving the user a single mechanism to access a wide array of existing services.
In a parallel to the Internet, information resources within SAS Institute were being made available with a growing diversity of tools. The Institute was facing information access similar to the ones that applied to the Internet, yet few people recognized the parallel. The current SWW administrator had the insight to identify that a similar solution could work at a local level: the same WWW technology could be brought to the internal network to produce a company-wide web. The almost hypnotic effect of the Mosaic browser made it the perfect choice as the initial tool to draw people into using this local web. At the same time, the Mosaic client was beginning to receive a lot of press, and many companies began to use the World Wide Web as a tool to provide information to current or potential customers. After experiencing the technology first hand for a few months with the local web, SAS Institute officials recognized the opportunity to provide customer service in this way, as well. As a result, two new and related projects were begun: one to provide internal information to employees, and another to provide information to current and potential customers on the Internet.
The initial SAS Wide Web support was very basic. After the client and server software was downloaded from NCSA, a simple example web was built to show how Mosaic could be used as an interface to multiple information sources. The NCSA HTTP server was installed with a rudimentary page consisting of links to the existing Gopher server, the News server, and the newly-downloaded documentation on the Mosaic viewer and HTTP server. It was a good beginning, but it was not enough to convince people that the SWW would be anything more than another tool they would have to learn to use. More value had to be added to the SWW, and the wealth of the Internet provided the solution: more tools were downloaded to furnish even more information from different sources. Rosetta Man was added to allow access to UNIX man pages, and texinfo2html was added to convert several manuals for GNU software (GNU emacs, gcc, grep, and so on). At this point, there were enough resources to test the waters and let a well-chosen subset of the SAS user community try to do something practical with the SWW. The software was announced to SAS Institute's internal tools development department and to the manager of the in-house debugger development environment. These people were selected because they were accustomed to dealing with start-up environments, and they would readily take personal responsibility for learning, exploring, and debugging a new tool.
At this point, the SAS Wide Web was a functioning information delivery tool, albeit a small one. It had simple interfaces to several sources and a set of live data, all accessible with one tool. It was time to open up the access, so the SWW was announced company wide. The user base grew, and the web administrator got more and more questions about how to link this web into other local tools. As a direct result of user dialog, an interface to the Staff Location and Information Monitor (SLIM) tool was created. This project was significant because it demonstrated concretely that gateways to retrieve information from other tools could be built successfully, and they could integrate elegantly into a single user (SWW) interface. One final aspect of the SWW needed to be addressed, namely setting up standard file locations for SWW development. The development environment needed to be familiar to encourage people to experiment. As a result, the web administrator built the standard SWW directories as an extension of the existing departmental directories. Because of this set up and because of the trail blazing by the initial users within their own departments, the activity in the web grew suddenly and exponentially. Access on the server went from one hundred document requests a week to over forty thousand accesses per week in a period of less than six months.
In order to continue to make the SWW available to as many people as possible, the web administrator also installed clients for other platforms at the Institute. Client software was downloaded for Windows (3.1 and NT), MacOS version 7, and CMS. Lynx was brought in to provide access for users with home terminals that did not have graphics capability. Besides giving access to a wider audience, having diverse viewers helps users distinguish the conceptual difference between the client and the SWW.
During this whole evolutionary process, more and more people attempted to write HTML files for the SWW, but most of the information about HTML was available via the Internet WWW. Most employees do not have direct Internet access, so local copies of the pages from the Internet were brought into a local directory, and the clients were modified to look in this directory when a user asked for an Internet file. This gave people the "feel" of Internet access without actually having to open up access. Another avenue for authoring was to use converters for word processing systems already being used at SAS Institute. Programs to convert from Frame, WordPerfect, RTF, LaTeX, and a few other formats to HTML were acquired and fine-tuned.
The SWW grew in popularity in much the same way as the WWW. When the administrator first set up the environment and announced it to the Institute, it was generally greeted with indifference, except for a handful of interested people scattered throughout the Institute. These people had, at one time or another, expressed the need for online documentation in a user-friendly interface, and when the SWW became available, they were enthusiastic about the SWW's potential in their particular jobs. As the first users in a new environment, they had the opportunity to direct the SWW's development significantly as they and the administrator became more knowledgeable. The importance of the first group of users cannot be understated; their enthusiasm, their ability to lay groundwork for future use, their willingness to experiment, and their ability to communicate the concept of online information delivery to others all played an essential role in priming the SWW for success. They also took responsibility for demonstrating, teaching, answering questions, and developing instructional documents on the SWW. The early users rapidly became the experts as more and more people connected to the web.
To encourage people to develop for the new SWW environment, easy methods for bringing existing documentation onto the web are essential. While most users find HTML easy to write, marking up text in HTML manually is considered a cumbersome task. If the only way to make a document viewable on the SWW were to edit the markup manually, most information providers would not make the effort. Fortunately, conversion tools solved this problem for many documents that needed to be accessible online. Much of the internal documentation at SAS Institute is stored in ASCII files, marked up in an experimental language developed at the Institute called pml. The pml tag set is not SGML compliant, and it allows richer description of text presentation than HTML. However, pml can map to HTML reasonably well, and a Perl conversion program called pub2html, was developed for this purpose.
One-time conversions are not sufficient for documentation in progress since it is frequently modified in its original storage format. For these documents, the original format must be the primary format with HTML secondary because HTML is neither suitable for hard copy book production nor for printed documentation. Yet, the most recent version of the documentation needs to be available online. The writer is then faced with having to run pub2html on every update to the original file, thereby maintaining two identical versions of the text. To solve this problem, another tool was developed to work in tandem with pub2html, using a script. The script runs pub2html "on the fly" when an anchor is activated; it pulls in the latest version of the pml file and dynamically creates an HTML counterpart. The author simply must encode the anchor's reference as the script with the name of the original pml file passed in as an argument. With these two conversion tools, approximately 80% of internal documentation can be made available on the SWW with minimal effort. To date, only a fraction actually is online, yet those that have been converted are among the most referenced documents at the Institute.
Similar tools have been made available for conversions and other useful niche applications. SAS Institute programmers have developed some of these, while others have been downloaded from Internet resources. The SAS conversion toolkit includes programs that take RTF (rich text format), plain text, man pages, newsgroup articles, CBT (computer based training) files, and internal database formats as input. Other utilities have been created for a variety of purposes, such as a conversion program that reads the user's online appointment calendar and creates an HTML version of the person's schedule for the next five work days. Because of the openness of the SWW, users can become developers when they discover a place where functionality can be added. Initially only the SWW experts developed tools, but now others have stepped forward with contributions.
Tools have made a great difference in the progress and perception of the web around the Institute. Rather than being just another application that requires a long time to master and has a lengthy and tedious conversion effort, the SWW has become an true integrator of diverse information. However, the true and unstated driving force behind the success of the SWW is, simply, fun. People like to create striking displays, to incorporate interesting images, and to experiment with hypertext. Mosaic, with its useful features and inviting interface, is fun and easy to use, and for these reasons, people are especially willing to use the SWW. Because they almost consider it play, people are likely to spend time exploring, to become aware of what information is available, and subsequently they find resources that help them in their jobs. With such positive underlying energy, the web has evolved in such a way to ensure its success. The idea of online information delivery has come of age at SAS Institute, and the SWW now makes it possible with ease.
As the previous section indicated, the SAS Wide Web has become widely popular at the Institute, and its usage is continually expanding. The tools facilitate making documentation available on the SWW, and individuals as well as groups have taken advantage of them to make information public. Open access is good for internal documentation, but not in all cases. Providers must be well aware that the information made accessible to the web is not protected in any way, and anyone can access it. Accordingly, no sensitive documents are to be put on the SWW; only information that anyone at the Institute can or should be able to access freely as a part of their jobs should be on the SWW. Generally, the type of information available on the SWW is support documentation of various types which one may need for quick reference in daily work.
Most divisions and departments at SAS Institute have established an online version of their organizational charts. These pages list the people that work in departments, divisions, and project teams, and often describe the responsibilities of the group. Generally structured in a hierarchy, they mirror the "reports-to" relationships between people, groups, departments, and divisions. When an employee is new or begins to work on new project with a different group of people, the person can easily get some background information about the unfamiliar organization. To fill in the detail behind organizations, individuals create personal home pages, where they usually list their office location, phone number, email address, and brief descriptions of their projects and responsibilities. Creativity, within practical limits, is also encouraged, and some personal home pages have become useful resources and reference points for many others. Building a personal home page is an excellent way for people to enter the SWW; not only does it add to the pool of useful organizational information, it also helps people learn HTML and become familiar with authoring online documentation. Furthermore, once they are given the chance to participate, people get excited and enthusiastic about the SWW and often go on to make new and innovative contributions to the SWW. For these reasons, putting organizational information on the SWW has been a worthwhile use of resources.
A vast amount of technical information has been made available on the SAS Wide Web, and the greatest benefits have been seen in this domain. In general, technical information falls into four categories: internal documentation, the Institute calendar, downloaded resources, and project tools. First, internal documentation includes technical manuals on software tools and the standards and guidelines for software product development. The SAS Wide Web currently offers documentation on the interactive C programming language debugger, source management tools, program defect tracking software, interface development guidelines, C programming standards, and the SAS internal programming guide. This documentation is ideal for online delivery because it is used widely throughout the Institute, it is updated and enhanced regularly, it is easily broken into contained topic units (which naturally translate to SWW pages), and it translates to HTML with minimal effort using conversion tools. The authors no longer have to be as concerned about delivering hardcopy to the appropriate audiences and notifying users of updates. Rather, after setting up the SWW access pages and scripts, they just maintain the original files as usual and let the users access it as needed. The users seem to prefer the online delivery as well. They can usually locate the information they need quickly with Mosaic's navigation facilities and with links to find related topics. In addition, they can simply access the manuals in another window on their workstations as they work. It eliminates a bothersome context switch to a paper document, which might have to be tracked down, borrowed, and then returned.
Second, the Institute calendar includes the software product release schedule that governs development timelines and schedule for the annual SAS(r) software users conference (SUGI). Obviously people need to find out the latest decisions about the product delivery schedules in order to plan their work, so it is natural to have it on the SWW. The users group conference information is also helpful, as a significant number of SAS Institute employees attend the conference. Knowing the dates and events well in advance helps in planning.
Downloaded resources provide a small, yet rich, online library of diverse information for local access. At the Institute, most employees do not have full Internet access so they cannot get to WWW resources directly. Often these resources are needed for daily work, such as developing HTML files for the SWW. The largest percentage of downloaded files has been help files and documentation on Mosaic, HTML, and the WWW protocols from CERN and NCSA. Other useful public documents are available as well, such as the TEI (Text Electronic Interchange) standards, several HTML style guidelines, Perl documentation, and the manuals for an HTML editor for the Macintosh.
Finally, the SWW enables employees to improve their work processes using online information delivery. In one application, a technical writer set up user documentation files for a particular product to be accessible on the SWW using the automatic HTML generation tool, pub2html. This effort is a first attempt to use the web as an online review application. The documentation is linked directly to the developers' pages on the software product itself. Anyone interested in this particular product need only refer to one small collection of pages to get any level of information about it; from technical specifications, to project plans, to user documentation. From the writer's point of view, circulation of many hard copies for casual and intermediate review is no longer needed; the developers can keep up to date on the progress online. Unfortunately, the viewer technology (Mosaic, in this case) is not yet capable of full online review and editing, but this technical writer has found an innovative way to use the technology that is available.
Another project support effort uses forms as an survey mechanism for product evaluation. In a project to perform acceptance testing on several software packages the Institute was considering for purchase, online forms in the SWW were developed to accept input from the individual testers. Formerly, such testers would record their feedback on a paper form (or in an ASCII file to be printed out) while they tested, which would then be manually keyed into an evaluation program. By putting the form online in the SWW, the process is improved by allowing the testers to input their evaluations online as they work in another window at their workstations. Errors because of rekeying input are reduced and the input is kept electronically throughout the process. As this, and all of the previous examples illustrate, a local web can, in fact, be used for a variety of purposes. This section has shown, through the examples in the SAS Wide Web, the great potential for a local web as a tool for dissemination of internal documentation and for development of web applications that improve work processes.
The SWW has become well established in the several months since the technology was introduced at SAS Institute. At this point, the future is practically assured, as the SWW grows in content and functionality, and the usage continues to increase. It is already evident that as more people become familiar with the technology, they will discover creative applications for it, in terms of interfacing with existing technology and adding features for specific purposes. Such efforts are already being given considerable thought, and some are underway.
The immediate future direction for the SWW is to make even more information available online and to expand the user base to include as many employees as possible. Fortunately, as the previous sections indicate, the climate at the Institute is favorable for continued development of networked information. Departments see the value of putting their public documentation online, and since existing tools simplify this effort, managers are willing to support this type of work. The challenge is to get others to think broadly about the SWW so that they may explore new ways to communicate publicly online. For example, weekly sales reports, which are usually cascaded through most of the Institute via email, are excellent candidates for delivery online. The information base will also continue to broaden in the future, as more conversion programs are developed worldwide. Already, SAS Institute has come up against the need to convert WinHelp, OS/2(r) information, and Quark, and the tools have not been available.
Another way the information base of the SWW will be expanded is by developing interfaces to existing resources. This initiative is in the same vein as the development of the pub2html script: translating existing information to the online paradigm with minimal effort. An independent SAS software product called InfoTap(r) is the first candidate for SWW interface development. InfoTap is a full-text search engine for collections of semi-formatted and unformatted files. A significant amount of internal information is already converted to InfoTap, and rather than duplicate the InfoTap database, the SWW will use scripts to access InfoTap resources dynamically via its gateway. Other possibilities along these lines are being considered for the future with emphasis on dynamic access of the SAS System, both SAS datasets and SAS programs.
The SWW can also improve by extending the online document review functionality in the Mosaic client. Writers, editors, and technical specialists have needed such a tool for a long time, and Mosaic offers a partial solution with potential. Mosaic delivers documents easily and clearly, and the SWW enables the documents to be connected to the entire realm of metadata related to a document, such as organizational documents and project plans. Once a writer has put a document online and a reviewer reads it, the next step, adding the actual review comments, should be the next step in this paradigm. But the tools do not exist. Ideally, online review teams need group readable annotations, location-specific annotations, electronic mail notification of review progress and completion, and integration with any other workflow and document management tools. Developers at SAS Institute are beginning to consider building Mosaic enhancements along these lines.
The key to the continuation of the SWW, however, is the availability of the free WWW software from NCSA and CERN. Should these software tools become commodities, the improvement of the SWW, like many Internet servers, will either stagnate, or rely on unofficial mechanisms for upgrades and fixes. While commercial viewers may become available, the SWW will depend on free software to survive.
As an information integration tool, the SAS Wide Web has become a valuable asset in its short six months of existence at SAS Institute. It has significantly improved internal information access methods because developing documents for the web is easy, local tools for document conversion allow existing documents to be put on the web without great effort, and its interface, Mosaic, makes the web a pleasure to use. During the course of this project, the SWW evolved rapidly for two reasons: a good integrated access method was desperately needed, and the technology was engaging to use, thus drawing a larger audience in. The result of these driving forces is a rich resource of information from many different areas of the Institute and the continued support for further SWW development. For the time being, a viable solution to the problem of internal information delivery has been found, using the WWW model and its freely available software and the creative extension of these technologies.
Lauren A. Bednarcyk is an applications developer and project leader in the Publications Division of SAS Institute Inc. She has a Masters of Science in computer science from North Carolina State University with a concentration in database technology and a minor in computer engineering. Her current research and work activities include object-oriented databases for SGML document management, writing for non-linear online presentation, and designing effective WWW documents.
Kevin D. Bond is a systems developer for SAS Institute Inc. in the Publications Division. In the past, he has been responsible for maintaining the parser and preprocessor for the SAS/C(r) Compiler, and now he is the web administrator for the Institute. His current research and professional interest is online information delivery.