Hypertext links in cyberspace are the subject of fascinating research being undertaken by Steve Hitchcock and Les Carr at Southampton University . They have been funded by  to explore, in the Open Journal Project, the use of links within the context of electronic scholarly journals. This work builds upon work being undertaken by the same team in a separate project, the Distributed Link Service (DLS)  . This is creating a system which separates the link data from the document and stores it in link databases or 'linkbases'. Sets of links can then be selected from the database and overlaid on the document being viewed. Les Carr, Project Manager, explained how this technology was being applied in the Open Journal Project. Storing documents and databases of hypertext links separately until the user combines them allows the user to choose different sets of links to appear in any one document. Hitchcock calls it 'link publishing at runtime'.
To show me how this works, they pointed a Web browser at an interface to the Distributed Link Service maintained at Southampton and Nottingham Universities. This allows the user to combine linkbases with other Web files, choosing the levels of links required. For example, a user may choose to apply a linkbase specific to a certain page, (which is the usual choice) or a 'context' linkbase which sets up links appropriate to a certain subject. Pages then displayed will contain hypertext links inserted 'on the fly' from the particular linkbase or bases chosen.
Watching this happen was simply stunning. First, Hitchcock and Carr ran some 'normal' pages in Web browser. Pages were displayed with as few or as many links as they had in their 'native state'. They then asked the DLS server to apply some linkbases and viewed the pages again. Documents which had had few or no links in them previously now positively sprouted with them. For example, a page about computing that had been set to display using a context linkbase relevant to UNIX now contained many links to pages elsewhere providing information about UNIX terms. The same document would display a totally different set of links if it was loaded using a context linkbase of terms relevant to, say, cognitive sciences.
Carr described the way the project aims to take a collection of archives of journal articles (both from the 'core journals' upon which a particular Open Journal is based, and related journals) and possibly of other electronic sources. These are then integrated by means of hypertext links. The information in any particular journal article is linked to all other information in the collection to which it is related by means of a linkbase. In this way, a journal is 'opened' out.
Again, a demonstration brought this to life. Displaying an article from one of the experimental journal archives with the relevant linkbase switched on showed words defined as keywords relevant to that subject area highlighted as links. Clicking on one of the links delivered other pages containing the word within the collection of resources encompassed by this particular 'Open Journal'. As Hitchcock and Carr recognise, the process of deciding which words should be defined as links is one that requires careful work. In Carr's words, "a link is not just an instruction to jump to another location, but an expression of a relationship between two documents".
We discussed the advantages of this approach. First, separating hypertext links into discrete databases aids the process of maintaining and updating them. Second, it allows one link to point to several documents - there are different display options to make it clear to the user that this is happening. Third, as described above, it enables different links to be superimposed on the same document in different situations. Carr speaks of 'information re-use'. "You can publish things used in one context one day, in another context another day". As Hitchcock explains, it functions as a direct way of giving people access to related documents. Searching using a Web search engine would be an alternative approach, but the link service gives direct access to related documents rather than requiring users to go down a Yahoo-style hierarchy . "The link server tries to project that information into the document you're reading at the moment".
In addition, the Open Journal Project, by linking the text, or at least the abstracts, of many documents from different sources, and by making a document available on-screen from a click within another document, tackles what Hitchcock describes as "the traditional library's lack of integration". Many categories of related material can not only be displayed in one place, but they can be displayed fast (or, at least, as fast as the Web allows). "Quality matters, yes, but speed does as well," commented Carr. "You can follow these things through with almost zero cost in terms of time."
Hitchcock, who came to the project from a background in publishing, illustrates the way in which Open Journals is wider than just an electronic journals project. The intention is for it to "act as a framework for on-line publishing". Linkbases can be published as quality products, with the links being guaranteed as relevant and reliable. "The value added element is contained in the links." Publishers might choose to maintain some linkbases as subscription products, particularly those relevant to their existing publishing interests.
This is an exciting project. "The single distinctive thing about the Web is links," said Carr. "A link service is at the heart of what it is all about." Watching the processes of integrating Web documents with linkbases, and of moving between the different pages within an Open Journal framework, bring out the potential of the Web in a way that is truly astonishing.
 Open Journals Project,
 Southampton University Web site,
 eLib (Electronic Libraries Programme) Web pages,
 Distributed Link Service details,
 Yahoo UK Web Site,