5. The 7 Design Concepts for Navigation in Cyberspace
"We commonly mistake data for information. Information starts with data, but data is not information -- it is a source of information."
--Ramesh Jain. [Jai95]
The general problem we would like to address in this and the following seven chapters is how hypermedia documents should be structured and what navigation mechanisms should be provided such that readers can orient themselves in large information spaces. These large information spaces can consist of free-text articles, structured databases, hypertext documents, or knowledge bases. The orientation and information exploration problem can easily be experienced by anybody browsing on the web: How often in a browsing session did the reader incidentally stumble upon an interesting site, only to desperately search the same site in the next session. Also called the problem of being lost in hyperspace, this question has found high attention in the hypertext community, and many different solutions have been proposed.
We have identified seven concepts that need to be considered when designing the structure of a hypermedia document. In the following seven chapters we discuss in detail these seven concepts and their application for the design of usable hypermedia documents. We illustrate the basic principles by a large collection of practical examples. We are therefore simultaneously using the seven design concepts to classify existing tools for navigation and information structuring into categories.
Frisse suggests a classification [Fri90] of navigation tools for hypertext documents distinguishing between signposts, shallow guides[4], and deep guides. Signposts are the menus and buttons that indicate the existence of a link and allow the user to follow this link. They may, e.g., be represented by electronic bookmarks. Shallow guides are overview maps and index lists that indicate the structure of the hypertext document. Deep guides extend the notion of shallow guides by incorporating knowledge not only about the structure of the hypertext network, but also about the semantic contents of the hypertext document.
While Frisse is the first one who has come up with a classification, we think that his taxonomy needs further refinement. We are suggesting a more elaborate classification of navigation tools into seven categories (see figure I.21). At the same time, these seven categories serve as design principles for designing usable hypermedia documents. The seven design concepts cover a broad range of navigation tools and techniques ranging from appropriate structuring of information to the application of artificial intelligence techniques [5].
Figure I.21 The seven design concepts for navigation in cyberspace
The linking structure is the most distinguishing feature of a hypertext document. Links allow direct access to a designated location within the information space through markers that are embedded into the documents. We distinguish between straightforward "static" links, and context specific, automatically generated "dynamic" links.
Providing searching capabilities is an obvious means for locating information. In hypertext documents, this is mostly done using full-text search, although there are also systems that provide additional databases for searching. In this book we will limit the discussion to full-text search tools mostly for the web.
For novice or first-time users it can be very helpful to reduce the complexity of the n-dimensional hypertext document to one dimension by offering them a sequential path or guided tour through the hyperdocument.
A hierarchical document structure is very well understood by humans. Almost all printed books are organized hierarchically. Prospective authors are well advised to employ a hierarchical document structure for a new hyperdocument, but there are also tools like Superbook that compute a hierarchical map from already existing documents. The hierarchical information structure is then made obvious to the users and put at their disposal as principal navigation aid.
Similarity links connect nodes that have similar contents, but might not yet be linked. An index is a very simple means to discover similarities between different nodes, because pages that have common index entries might exhibit some sort of similarity. More complex tools in this category are based upon the assumption that the system not only has knowledge about the document structure, but also about the contents of the information contained in the document. There are two approaches to this problem. IR-based systems compute similarities based on statistical approaches, while knowledge-based systems try to use a deeper understanding of the semantic contents of a document. Obviously, this later requirement is very hard to accomplish and needs some sort of expert system. It is therefore not surprising that the systems of this category have not overcome the state of early prototypes.
Mapping is a technology to structure, visualize, and manage webs of information. Similar to real maps, graphical maps show readers where they are, where they can go from here, and give them an overview of their local and global context. They are thus one of the most flexible, versatile, and user friendly means for navigation in cyberspace. Mapping is orthogonal to the previous concepts, in that maps can be used to visualize links, search results, sequential paths, hierarchies, and similarities.
The concept of the intelligent agent is a very popular one not only for navigation. The agent metaphor is well understood by humans, because agents simulate human assistants. The systems in this category incorporate artificial intelligence-based techniques derived from the metaphor of agents assisting human readers in their complex orientation tasks. The agents are implemented in different ways ranging from simple, hardwired guides to rule-based agents that are able to react more flexibly to different needs of different users. Agents can use any of the previously discussed six concepts to assists users in their navigation task.
Most of the tools and techniques described in the following chapters have explicitly been developed for hypermedia documents. With the advent of the web, the distinction between hypertext and sequential documents has been blurring, as many existing, sequential documents have been converted to hypertext. It is one of the goals of this book to have authors not only change the format from, say, Microsoft Word to HTML, when converting a document from text to hypertext, but to specifically redesign the document for presentation in another, much richer, environment.
Chapters 6 to 12 now discuss each design concept for structuring and accessing information by first describing the problem in its context, and then presenting a broad range of possible approaches to give the reader a range of tools and techniques to implement the concept.