Linear versus Non-Linear Systems

Hypertext has not always been the default way to engage readers. Before this was the preferred method, we had linear text. Linear text only goes on way: from the beginning to the end. You read the information as it is listed on the page, the way the author intended for us to read it. In this case, keywords and phrases are irrelevant since there is no reason to skip around. In some cases, this is still a method that makes sense, such as a mystery text or website that you wouldn’t want to spoil the ending. There is a heavier focus on keeping the reader entertained since they have to make it through the entire document in order to gather the needed information. Some readers enjoy the linear style because it is the traditional method of reading and learning, but there are benefits to both styles.


A survey done by Davida Charney entitled, “Enhancing Hypertext Documents to Support Learning from Text,” discusses the tendencies of readers in 1992. Their hypothesis was that readers could possibly get lost in hyperspace because it is often confusing and hard to follow. They produced a study to see which of the five different hypertext versions are the most beneficial for users. These options all had the same content but were organized differently. The options are: free network structure without map, free network structure with map, constrained structure without map, constrained structure with map and a control version that was a standard linear format. The free network structure is the hypertext version, with the constrained structure being the linear version. From the results, the group without a map in the linear category did the best in the question probe and problem solving sections. In the subjective ratings, regarding how much users liked the system they were using, users like the constrained system over the linear. They said they found the free-network system often lead them astray and they would get stuck. This study is of interest because it shows that readers tend retain the information best when it is provided in a linear format. The study notes that this is “undoubtedly because they were ‘encouraged’ to read all of the screens” (Charney). The study concluded by telling us that if it is critical readers learn certain material, we cannot rely on them to click through to find it. If you generally just want students to browse around and peruse for information, a hybrid form of the constrained structure could work well, as students seem to enjoy the constrained system better. Overall, this study provided some background information on how students feel and work with these systems, both linear and nonlinear.


Expert Systems

The main goal of hypertext is to create and effective and functional platform for users. As the creator of such documents, it is a challenge for us to not control the experience of the user too much. We want them to be able to make their way around, but not restrict their freedom too much while browsing. Expert systems are a way of increasing the usability of a hypertext document, and there are two specific kinds: example-based systems and rule-based expert systems. The knowledge engineer is a specialist in the field that uses these systems to document how experts make successful decisions in the field. In example-based systems, they focus on documenting important details in case studies and uses inductive reasoning from these studies in order to solve problems. The more examples made available to the system, the more reliable the results. It can learn to generate rules and/or generalizations from the case studies given and takes the load off the knowledge engineer. In the rule-based expert system, the same knowledge engineer creates an “if-then” scenario, called a rule, to represent the knowledge of the experts. One example of this for an online store could be IF (the car is smoking) AND (your oil light is on) THEN (your car may be leaking oil). This is a simple example, but is a piece of knowledge an expert would need to solve the issue at hand. This is one piece of the collection of rules made to diagnose auto problems, often called a knowledge base.


These expert systems consist of the knowledge of many experts in the field, and as previously mentioned, the more knowledge the more effective. Within these systems is something called an inference engine, an engine that decides which rules it needs to apply, and in what order they need to go in. Sometimes, the system needs feedback from the user to better answer the question at hand. In the example above, it would presumably ask if the car was running or what the issue was in order to continue along. If the user isn’t available or doesn’t have the answer, the system will continue to draw conclusions based on previously gathered data.


Using an expert system to help with a management, control or diagnosis function within a hypertext can greatly increase the usability. There are three things that expert systems excel at in reference to controlling a hypertext application, these are: “Integrating with other applications, solving complex problems nonlinearly, and working within clearly defined boundaries” (Glover 630). These expert systems help increase usability in hypertext because they provide the means to do a run-time audience analysis/needs assessment, which helps tailor documents to fit the users requirements, and then they can advise said user on how to use the document.


This brings us to rules within the hypertext. “Rules can control what parts of a document are accessible at any point in a session by providing progressive disclosure based upon initial session goals and the user's path through the document” (Glover 630). There are three kinds of rules that may be integrated in a hypertext, these are: front-end rule bases, concurrent rule bases and back-end rule bases. The front-end rule base is for the start of the document, where rules can conduct an interview to collect information about the users needs. The concurrent rule base exists when the user has some sort of history with the topic and the system can begin to infer about the users preferences as they move along. Obviously this doesn’t work it they are browsing at random. The back-end rule base is described as a sort of exit interview with the user to find out if their needs were satisfied as they went though the document. If we can find out where the system succeeded and failed, the information and data collection can go back into the database for analysis by the rules for the next time they visit the document.


There are a few other ways that expert systems can help provide customized user support by adapting hypertext documents, one of those ways is by providing a customized hypertext structure. Because there are many reasons a user can come to a document, by having rules at the front-end of the hypertext application, the system can customize topics and links to create a custom document to answer the users needs. If these needs happen to change as they move through the hypertext, the concurrent rule base can add and remove links as necessary. Another way these systems help users attain the information they need is by implementing progressive disclosure. Some documents require us to use a more linear system, say a how-to guide or a training manual of sorts where the user needs to be walked through the process. In such cases, technical communicators say it wouldn’t benefit the user to allow them to skip ahead and that too much freedom in this sense isn’t beneficial. This is where the rules can help control the pieces of the document that are available, and the progressive disclosure can protects the user’s path.


It is no doubt these expert systems can benefit users, and some of these rule based systems help better meet their needs. “In general, expert systems provide a great deal of flexibility and control. And because rule-based systems are nonprocedural, they tend to be simple to maintain; in most cases, rules may easily be added or deleted as needed” (Glover 634). This, in addition to the expertise brought to the hypertext from the knowledge of the author can really benefit the users and help them achieve whatever goals them came with. Not all users need to be walked through a document at a slow pace, so this does not necessarily apply all of the time. “Despite the capabilities described here, however, expert systems cannot replace good document design, effective navigational devices, and clear writing,” (Glover 634) which is where the technical communicator shows its benefits to hypertext again.


Next: Rhetoric & Thinking