Root maps to missing links

The metaphorical world of bytes and wires did not exactly obsess Wendy Hall as a teenager with a gift for problem solving. Her awareness of computing capabilities was not matched by a burning desire to punch card and argue the finer points of code. It took a chance encounter with an early-version microcomputer some years later to change that.

Now, as one of the handful of women in senior positions within academic computing, Professor Hall leads the Multimedia Research Group at Southampton University hoping to change the way the World-Wide Web works, using Microcosm multimedia information management systems.

Professor Hall's first calling was in mathematics. She got a first at Southampton and stayed there to complete her PhD on non-Euclidean crystallographic groups and Klein surfaces. "I was always good at maths. It came so easily for me and I enjoyed it. I could complete my homework very quickly.

"I always assumed I would teach and, in fact, I hated my early experiences with computers using Fortran and punch cards."

But as a maths lecturer teaching BEd students at La Sainte Union College of Higher Education, she used a Commodore PET microcomputer for the first time in 1978, taught herself BASIC and over the years migrated from that platform to interactive videodisc and HyperCard.

"I've been lucky because I was involved in videodisc applications early on and could see the possibilities for blending text, images and sound.

"But the technology was not portable enough and was also a little unreliable."

Now the technology has caught up with the vision and Professor Hall is able to realise the projects first seeded in that transitional period, applying sophisticated techniques to create linked multimedia information systems.

Out of that culture has come the project, now five years grown and still developing as an application for academic and commercial use. Microcosm is a hypermedia system which allows users to browse and query large multimedia collections.

However, Microcosm differs significantly from most hypermedia systems through its ability to integrate information produced using a variety of third-party applications. Microcosm acts as an "umbrella" environment in which the user makes links from documents in one application to documents in another. This also allows access to data held on read-only media such as CD-Rom and videodisc. The project's latest development aims to alter fundamentally the way people access information on the Web, removing the barriers to quick retrieval of precise data about given topics.

At the moment, the Web is a fascinating and infuriating creation - offering the potential of unlimited information but with limited means of acquisition - hypertext links and more recently, search engines.

Professor Hall explains: "What we have with Microcosm is a complete information management system, applied globally on the Web. It is more than just hypertext. This will change the way we work on the Web but it won't be painless. And whether in the long run we get the code accepted by the Internet community is less important than having the ideas in place."

That said, the response from the Internet community has been encouraging and the project group are considering the option of joining WebCore, the European R&D arm of the www.

"At the moment, the Web is a pain. It is limited, it is difficult to find exactly what you want and is incredibly chaotic. Tim (Berners-Lee) has created a wonderful tool and knows about our application. We all realise we need to add to Microcosm's local network concept to make it global now. The limitations of the hypertext model on the Web are well-recognised and we know that a link service is needed."

She explains: "Designers of hypermedia systems have recognised the need to move away from closed systems to open environments which separate the link structure from the data in the system, and enable separate link and data processing."

This work, she explains, is essential if the Web is to develop. It should reduce the authoring effort in large-scale hypermedia applications while making them easier to customise, modify and extend.

Professor Hall's personal space seems regularly modifiable - she manages to slot in a 60 to 70-hour week somehow. But the "online 24/7" mentality (let's hook up to the computer 24 hours a day, seven days a week) leaves her distinctly cool. A Mac user by culture, she has a PowerPC 7100 and a portable Powerbook at home but a Toshiba 486 laptop as her constant work companion. "I enjoy my work very much but I also try very hard to keep Friday night to Sunday lunchtime free. I get up to 100 messages daily on email but I do not have them redirected to my home mailbox. I can pick them up if needed. So much of email is administration and I have to do that at home; I'd rather leave it for when I'm at work. We live in the New Forest and walk a lot there. It is ideal for getting away from connections. No one can use a mobile phone there!" And for an inveterate problem-solver the aesthetic agonies of choice presented in the high streets and shopping palaces are an absorbing pastime, but she is more smitten by the mix and match of clothes than by "all that techie stuff", a task she happily leaves to systems administrators. It is coincidental, though, that one of her commercial research projects is to configure a system using Microcosm for Marks and Spencer.

She is also happy to leave the hard code details to those whom it pleases, concentrating on using new technologies as teaching and learning tools in the broadest sense, which is what attracted her to the Web in the first place. "But the Web as a hypermedia system is closed in the sense that the link information is hidden within a document's data," she explains. "This is not a fundamental design feature of the Web but comes from the way Web documents are created."

Web links lead outwards from one document to others. The power of the Web is that a document can point to another without any action from the creator of the target document. A weakness is that however much a document creator would like others to point readers towards their work, they cannot arrange this without the help of those people.

Microcosm provides a kind of reverse hypertext link, known as a generic relationship, which could solve this problem. This allows a user to associate a document, such as this article, with a keyword or phrase such as "Southampton University". Then any reader who encounters that phrase in another document can automatically retrieve this article. On the Web there is no such mechanism. There are search engines such as Lycos and Yahoo, but these index documents automatically. The Web provides no way for an author to declare what a document is about, or to indicate which other documents should link to it.

As Les Carr, senior research fellow and a key figure on Professor Hall's team explains, it is practicable to design a link service which adds value to the Web. But this extra software the "Distributed Link Service" must communicate across the network via a protocol. The Microcosm protocol is of a set of possible messages, each of which can be encapsulated into a packet of text containing all the information which an isolated piece of software would need to fulfil a request.

It is possible to publish sets of links as well as sets of documents. If your document mentions these topics then you can link to these places.

When a user fires up Microcosm on the network it will potentially be talking to dozens of document and link servers, and processing thousands of documents and hundreds of thousands of links between different media. Professor Hall and the Microcosm team are now working with the high performance computing research group at Southampton to tease out the processing power conundrums presented by such a gargantuan undertaking.

There are a number of plans to build on the project's linking facilities over the Internet. One, funded under the Realising Our Potential banner, is to build a prototype networked-Microcosm by this October. The emphasis is to analyse the protocols needed for an effective global documentation groupware system.

To encourage young women to pursue computing careers, Professor Hall worked with Gillian Lovegrove on the IT EQUATE project, which later blossomed at Staffordshire University (see page xii).But she herself has met few problems in male-dominated disciplines.

"I've never been worried about being a woman in this environment. I was the only woman mathematics PhD at Southampton during my studies and it made no difference. There were some definite disadvantages in the late 1970s and early 1980s. But I don't find any problems now. The computing world is undoubtedly very macho - I think it all began to go wrong when the microcomputer brought in a "shoot 'em up" games mentality.

"But the Internet is changing all that, but slowly. Girls who want to use computers can do so anonymously now. There are some very good code writers out there now who happen to be women but more could be done to encourage them into the area.

"Quite what would work is difficult to say. Judith Church's Women and IT bill provides some options. It may be that girls will have to be bribed into the system! There is no doubt that it would change the culture of computing for the better, a balance to the boys' games culture."

More information about Microcosm can be found on http://cosm.ecs.soton.ac.uk or mcm@ecs.soton.ac.uk. The commercial version of Microcosm is available from Multicosm Ltd at the same e-mail address or by telephone, 01703 595053.