Interests: Node-based Knowledgebases
Most reference or knowledge-based websites are constructed in a hierachial mould, with categories containing sub-categories. One of the drawbacks to this layout is the "dead-end syndrome" where a surfer finds the exact category they require, but there is no further paths to other relevant material - no suggested further reading. To move on from this, a surfer normally has to back-track, maybe even to the top of the hierachy before they can look at alternatives. This is a very limiting factor when discovering knowledge.
Another drawback of hierachial structured websites is that its only possible to see one view of the categories, from top-down. This leads to a situation where seemingly logical relationships are not visible to the user, which normally forces the visitor to go back to the top of an information hierachy and try again along a different route. A way around this, as Yahoo and DMOZ have demonstrated, is to have inter-branch links. This too is limiting - owing to its artificiality - that jumping across a hierachial tree isolates the visitor from their original path and thus losing the original thought process (and intentions) of the visitor. Get rid of this enforced hierachy and then we start having websites that mirror a users intention, rather than the guesses of a webdeveloper.
The World Wide Web is not a hierachial structure, and its that particular quirk that makes the value of it much more than the sum of its parts. My idea is to create a website that mimics the World Wide Web, while introducing some of the ideas of the proposed Semantic Web.
A typical navigation path of this type of website would consist of four ways of proceeding:
Relevant Links: Links to articles within the same topic, this is navigation within the current topic only. If the current topic is the one the visitor wanted, these links provide him with navigational routes to all the articles.
Refinement Links: Links to sub-topics of the current topic. This allows the user to refine his requirements, narrowing down his search to more specific articles than he currently has on the current topic.
Generalising Links: Links to parent topics of the current topic. Since the parent/child relationship here is a many-to-many relationship, topics can have two or more parent-topics. Being able to expand the search criteria avoids the "dead-end syndrome" without having to jump to the top of a hierachy of topics.
Global Function Links: Functions such as search, glossary, topic list, site map, discussion. This is the fallback when there isn't a logical relationship between the current topic and where the visitor wants to go.
The methodologies and techniques implemented within this sphere will form the basis of a webdeveloper's resource/portal, or a science-fiction related portal.