The web needs to get personal

In the 1990s, we dubbed the Internet the `information superhighway`. So why is it still so hard to find what we are looking for online? According to Prof. Wendy Hall of the University of Southampton, it is because the web is mostly linkless. What`s more, if we want the Web to be useful in our daily lives, web links will have to become much more personal.

Prof. Hall is head of Southampton`s Intelligence, Agents, Multimedia (IAM) Research Group. She says that hand-crafted websites generally contain few links because they are too difficult to maintain. `It is hard enough to maintain the content, let alone the links.`

Web links are also notoriously frustrating. Computers let us click on objects and text, but only to follow information paths that have already been chosen. Using current systems, users are destined to remain forever lost in cyberspace, because designers cannot anticipate each of the thousands of different ways that people might want to use – and develop – the information contained within their pages.

`Say I`m looking at information about beaches and that makes me think about ice-cream,` says Hall. `I might want information about where I can get ice-cream. Someone else looking at the same information might want to know about the best surfing beaches. Someone else might associate beaches with the D-day landings and will want information about the second World War. All of these associative links can`t be predicted by the designer of a web site.`

But what if links were not pre-determined? What if information paths could be dynamically created and tailored to our interests? Hall and her colleagues have developed new systems that do just this. The key to such systems are paths called `user trails`.

`User trails are basically a record of what information the user has visited,` explains Hall. `They can be automatically created by the system as a user navigates around an information space.`

The idea is that by having collections of “pre-recorded” trails that users have found useful, automated processes can be created to help others with similar interests find information in the same context. `The matching of users, the tasks they are trying to perform and pre-recorded trails can be undertaken by automated processes such as agents,` says Hall. `These agents ask questions such as “Who else has seen this document?” and “What other documents did they read?”. They work from the experience of others.`

Hall says travel guides use a similar principle in the non-digital world. `Someone has visited the city before and we use the information available in the guide to help us get to where we want to go,` she explains. `Printed travel guides aren`t so personalised but web-based ones could be.`

In the future, agents and user trails will help people learn by virtually following those with similar interests. `We won`t look for people who are going the same way as us, but our agents will,` says Hall.

Although the web is often presented as a mechanism that replaces human contact, Hall`s research suggests that the very opposite is true. Internet systems which rely on `user trails` will have human activity patterns integrated in their very core.

`At the moment we all see the same web. We actually need different webs, different sets of links, depending on what we are trying to find, what we know already and what context we are in,` says Hall.

The job for technologists, says Hall, is to put these more personal elements back into the system to reduce our dependence on pre-determined paths and links.

`One thing is certain,` says the scientist, `the Web will disappear behind the screens! And with it will go the buttons. Hypermedia is more than just point-to-point links. It is about making connections or establishing relationships. What we need to do is go “back to the future” in reducing our dependence on buttons.`