Even if hypermedia primarily is a tool for structuring and presenting information in a natural way for the user, problems may occur. Three problems especially recur: structuring, orientation and learning problems. Structuring problems may occur when an author wants to structure his thinking and ideas for hypermedia use. Orientation problems arise when a reader loses track of his whereabouts in the hyperdocument. Learning problems are those a user may encounter when starting new hypermedia documents, or when he is new to hypermedia.
In larger systems designed for information exchange among a number of persons, a possible problem area is the publishing process. As the publishing process in hypermedia normally is simpler than for traditional publishing of printed material, it appears that some users tend to publish their messages too soon. Another problem may arise when items are published which really should never have seen the light of day, for example something that contravenes the laws in force.
A number of different strategies have been drawn up to avoid or at least alleviate the problems dealt with in this lesson. The various methods largely build on existing features or extensions to existing multimedia features. Hence this lesson must be considered in the context of the preceding lessons.
Some of previously described hypermedia mechanisms in themselves reduce some of the problems. Database diagrams are examples of a mechanism which is a direct, important aid to user orientation in hypermedia. We will therefore refer to such mechanisms which have been described previously where necessary.
A number of problems may arise as a result of the special way hypermedia structures information. Conclin (Conclin 87, p. 40) mentions some of the problems hypermedia authors might encounter. Hypermedia requires an author to execute a number of actions while pursuing his or her own train of thought. He must, in the event of a new idea, construct a new link, construct a new node, and decide the text of the new node while at the same time keeping track of the idea behind the original node (the referring node). He must furthermore invent descriptive names both for the new node and the new link (if the link name is different from the node name). Finally, the user must determine whether the link is part of the document structure or whether it is a reference.
The reader may also encounter problems. Conclin points out that in linear media the author has decided on the order to be followed by the reader. But in hypermedia this choice is made by the reader himself. The reader must quickly determine which links to follow, and which links to skip over. The user may not always be able to do this adequately, and may therefore "lose the thread".
Some of these problems are typical of the writing process itself, for example managing the sequence of ideas behind the subject under processing. New ideas pop up before the author has managed to write down the previous ones, ideas are flowing - a fantastic sensation, by the way. These problems are thus not directly related to hypermedia. Some other types of problems may be reduced. Conclin mentions template nodes, cf. Lesson 6, section 2.2.5, as a means of reducing the number of operations required when constructing a new set of links and nodes. This may be carried out if the user has a set of available template nodes complete with link icons, rendering both node appearance and structure ready for use. This allows for a more consistent appearance because a fixed place is allocated to some of the contents of each node.
The author's problem of selecting the correct link word, link type etc., and the user's problem of selecting the correct link are naturally related. The author faces a number of pressing problems. He needs to express himself non-sequentially through hypermedia, he must create documents which the reader can understand and navigate in, and he must structure the information so that the reader considers the information received as relevant. It appears possible to reduce these problems both by means of mechanisms inside the hypersystem and factors external to the system.
Evenson, Reinfrank and Wulff (Evenson, Reinfrank, Wulff 89) have suggested ideas for a possible hypermedia design language, as described in Lesson 6, section 2.1.3.1. This includes standards for the appearance of the various types of link icons. If the author and his reader place the same interpretation on these icons, the user will probably be more easily able to decide which links he wants to follow. Thus some communication problems between the author and the reader will be reduced.
It is also important that the author himself devotes attention to constructing hyperdocuments which make it simple for a reader to follow. Landow (Landow (87) has proposed a set of rules for hypermedia authors, intended to ensure that the author considers as many aspects of a hyperdocument as possible. Landow deals with matters relating to link positioning and appearance. Moreover, he points out that the links must be significant to the reader, meaning that the reader must be able to rely on these links to bring him relevant information. Landow emphasises that the author must utilise the features offered by the hypersystem, for example database diagrams.
A related problem occurs when there is an overwhelming number of links emanating from a node. The user might become confused when he does not know which link to select. This has been reported among others by Vetter, Spell and Ward (Vetter, Spell, Ward 94, p. 56). Reducing the number of links will probably help the reader select the appropriate link.
Another problem of structure, mentioned by Halasz (Halasz 88, p. 845), is that at an early stage of the writing process the author may not be able to see the extent and structure of the hyperdocument under composition. This means that the author must be able to modify the document structure a number of times during the writing process. Halasz points out that only a few hypersystems permit changes in the structure of an existing hyperdocument. He therefore thinks that a requirement for new hypermedia systems should be that they allow modifications to existing hyperdocument structures. Furthermore, there might be a need to view different versions of the document.
The special structure introduced by hypermedia often causes a new problem for the user, i.e. difficulties in orienting himself in the hyperbase. This confusion often arises in large systems, typically in documents with more than 500 nodes, especially in unfamiliar documents with irregular structures (Halasz 88, p. 842). The problem is twofold: how does the structure of the document look, and how does one find the desired information (Conclin 87, p. 38).
It appears that most orientation problems can be alleviated using existing hypermedia features. Landow (Landow 87) writes, based on experiences using Intermedia, that disorientation problems have been reduced by means of text searches, document folders, links, diagrams and the option of following links via the main menu. Using these features, the user will be able to orient himself and/or locate relevant information.
Nonetheless it must be assumed that disorientation still might occur, in spite of the mentioned features which are available in hypermedia. Therefore, a number of techniques exist which are intended to further reduce the level of disorientation. Some of these appear to work, as described, among others, by Nielsen (Nielsen 90). The following sections describe some of these techniques. The Grimstad model is one example of a method which helps the user retain the overview, using metaphors, key screens and the accumulated experience.
The user commonly will have entered the base or part of it via a start node. Using a Home button, a user may at any time return to this node, the Home node. A user may employ this feature if following the links has not worked at all. If previous nodes were of interest, the "Back" option may be used.
The system tracks user movements, storing which nodes the user has visited. This enables a user to go back to preceding nodes. The user may thus untangle himself by going back to previous nodes until he recognises a place, allowing him to start again from there. One way that nodes may be stored is by using a stack for storing node identifiers.
Another version of the "back" option is allowing the user to return to any of the preceding nodes. HyperCard features the option of showing previously visited nodes in a compressed format, so-called "thumbnails" in a screen. This lets the user quickly see which nodes he has visited and navigate there with a double click. The Windows' help engine presents a list of previously visited nodes to the user, who may then navigate directly to one of these nodes.
Snapshots are created by freezing the base in its present state. The user may at a later stage return to a previous state in order to continue from there.
Personal paths allow the option of storing a sequence of nodes leading to information frequently accessed by a user. Employing these a user may follow an information path to where he knows he will find information.
The bookmark metaphor lends itself well for allowing the user to return to selected nodes. A user will then tag those nodes he may want to return to later. Some word processors use bookmarks too, notably WordPerfect 6.0 and Word 2.0/6.0/7.0
Landow mentions the landscape metaphor (Landow 87). Corresponding to a geographical landscape with special buildings, mountains and other landmarks, a hypermedia system may use particular nodes as "landmarks". A typical example of such a landmark node is the opening node of a hyperdocument. In other cases such nodes will mainly be constructed by the author, not being directly supported by the hypersystem. One way for an author to do this is to establish a number of fixed links in the nodes which lead to particular nodes, for example to a start node and a landmark node.
The possibility of keeping several nodes open at one time is essential for system orientation (Nielsen 90, Landow 87). The reason is that systems which permit only one node to be open at one time will cause user confusion, as the user will not have an overview of the information. This may mean that the reader does not really knowing how he came to be where he is and then that he does not know how to proceed.
Conclin (Conclin 87, p. 19) mentions standard windowing operations on each node, including the option of several open nodes at the same time, as a typical feature of hypermedia. However, some hypersystems do not feature the option of several nodes open at one time, for example KMS and applications based on HyperCard.
By tagging the nodes he has visited, a user will be able to see where he has been in the hyperdocument. Nielsen lists two different types of tags (Nielsen 90):
"Footprints" in the database diagram: Nodes already visited are tagged with a "footprint" in the diagram. This offers a user a quick overview of the hyperdocuments and where he has been.
Time stamps: Another way of showing that a person has been in a node before is to stamp the node itself with the time the user accessed it. If no time stamp is shown, the user has not been there before.
After a period of time tag validity will decrease, perhaps because the user is searching for other information. Nielsen therefore suggests that time stamps and footprints should be greyed out after some time, in the end to disappear.
The node name should preferably be unique and should also enable the user to associate something with it. By applying well thought out names to each node it will be easier for the user to understand what the on-screen node will provide.
If a user is to proceed through a system which in any way seeks to measure user performance, progress indicators are one way of showing this (Apple 89). These may be the number of possible points attained, the percentage of the total number of nodes or other markers which indicate how many nodes have been visited. Progress indicators may be used if the user does not need to be familiar with the application structure.