Everyone involved with scholarly journals has a problem with the Web. Readers wonder whether they will find what they want, and librarians want to bring some order to the Web's unregulated chaos. Authors are concerned with recognition, and publishers with how they can make money from the Web. The Web presents many new opportunities, especially for academic works. For those those who want to exploit the Web it is just possible that one approach, such as is being developed in the Open Journal project, could assist in providing a single solution to all of these problems.
Think of the Web as a car engine. In this analogy Web pages are the mechanical moving parts, and links are the engine oil. Without oil an engine will seize up. Web information is dynamic, always changing, so links have to be maintained more frequently than the average car service. The problem with the Web as it stands is that the links are treated as mechanical parts: link data has to be coded within documents. So adding links to documents and link maintenance are both more difficult than they ought to be.
The Open Journal project treats links as the oil, using a system, not uniquely, that separates the link data from the document and stores it in link databases, or linkbases. Sets of links can be selected from the database and overlaid on a document being viewed. The management of links and the superpositioning is handled by what is called a link service: we are using the Distributed Link Service, developed at Southampton University. Sounds difficult? Anyone can use the link service to follow and make links. What the user sees is either a forms interface or drop-down menus, so interaction with the service is controlled simply by a mouse or keyboard prompts.
That is the technical part. What the project is doing, since it is a publishing project, is to make this work for people, both users and publishers, and this is the difficult part. Applying links in this way is imposing a new publishing model, and people are sceptical, not surprisingly since there is no precedent.
The project's first open journal is in biology, a field particularly rich in online information. We are working with the Company of Biologists, Academic Press and Electronic Press, with its online BioMedNet service. The 'journal' has:
- a document management service, a sort of table of contents which is structured to cover a large range of resources;
- an interface to the link service, which allows users to customise the way in which they want the information to be presented;
- precompiled sets of links grouped into categories pointing to such resources as online journals, an online biology dictionary, molecular image databases, an abstracts database and so on. A user could select one or more of these categories.
Consider an example. A biologist is browsing the Web through the link service interface, by setting the proxy within a Web browser to point at a server hosting the link service. The biologist is viewing a technical paper with all the above link categories 'switched on'. Within the text the link service highlights a particular term: it looks like any other Web 'button'. The link could point to a number of documents - note, this could not happen with a conventional Web link - which would be listed should the link be followed, possibly including accessible journal articles that use the term as a keyword, say, or the entry for that word in the dictionary. If using a comprehensive abstracts database, every reference would be linked to the corresponding abstract, or to the full-text article if available. How often do you see journal papers with this number of links?
So on to further articles, with the same set of links applied to all articles viewed, whether linked from the previously viewed article or not. Every document visited, from whichever resource, is linked into a seamless web, whether or not the original document has any authored links.
Yes, people may be sceptical, but first indications are that this way of following information is going to be powerful for users. The project has a test user group. Feedback from users will be used to refine the interface and provide many more links.
Open journals are also planned in the areas of computer science and cognitive science. Collaborating publishers include the British Computer Society, Oxford University Press, John Wiley, MCB University Press, the Institute for Scientific Information (ISI), and we will also be working with publications in psychology edited by Stevan Harnad.
In due course the interface will be generally available on the Web for others to try. Access to subscription-based journal services cannot be guaranteed, but there is plenty of good information that is openly accessible.
Development of the link service interface, and of the links themselves, is managed within the Multimedia Research Group at Southampton University. Links can be applied to html documents and, because it is a popular format with publishers, to documents viewed with Adobe Acrobat. The capability to superimpose links is not available with the commercial Acrobat package, but a 'plug-in' has been developed for the project by the Electronic Publishing Research Group at Nottingham University, which has much experience of Acrobat publishing through its CAJUN project.
Perhaps solutions to the problems cited at the beginning are now becoming apparent. Links provided by a link service add a new layer of organization to the Web. For publishers the 'journal' now becomes a set of links: substitute links for the glue of the paper journal. A paper may be ordinarily available on the Web, in a preprint server say, but how can you choose to see only papers that have been refereed? Subscribe to a publisher's link service. For authors this gives accessibility; also recognition conferred by inclusion in a refereed source, just as with paper journals.
It adds up to quality information accessible at the desktop, which seems to be the current demand, with scope for commercial value-adding, the hallmark of good publishing.