Department of Electronics and Computer Science Southampton SO 17 1 BJ Navigating Hyperspaces with Contextual Structured Linking

In recent years hypermedia has been popularised by the rise and rise of the World Wide Web (WWW) [4]. This allows users to access distributed documents and navigate between them by following links, references embedded in the document currently being viewed. However, it has been argued that the WWW is nothing more then a distributed file system [47] and that its hypermedia functionality does not match up with that specified by early pioneers as being fundamental to a hypermedia system [8][21][43]. Open Hypermedia Systems (OHSs) as embodied by systems such as Microcosm [22], Chimera [1] and DHM [24], separate links from data, storing the link separately in a linkserver or hyperbase. This allows the system to monitor these links much more closely, in many cases guaranteeing link integrity, something that is impossible with the WWW [16]. It also allows for the possible querying or reasoning about these relationships [50] and the dynamic creation of links by the system [14]. Given these advantages it is curious why, if OHSs are technically superior, the WWW has enjoyed so much relative success? One possible answer is the way in which the WWW experiences network effects [55]. This is the property of some systems to become more valuable the larger the system becomes. A telephone is a classic example. A telephone becomes useful as soon as two people own one, buts its usefulness increases in proportion to the number of users. If the WWW is successful due to network effects, what feature enables it to exploit these effects, features that are evidently missing from OHSs? One possible answer is the strict use of open protocols across the WWW. Both in data format (HTML) and also in the communication protocol (HTTP). These standards allow any conformant client to gain access to the system. OHSs on the other hand, have traditionally used proprietary, or third party, clients that have been adapted to talk proprietary protocols. This means that a hypermedia client can only talk to one of many hypermedia systems. In addition, proprietary meta-data means that hyperstructure created 4 on one platform cannot be moved to another. In effect OHSs become information islands cut off from the rest of information space. Recent years have seen much effort being spent to create distributed, communicating OHSs [33][31]. However if these OHSs still do not interoperate with different systems then they remain information archipelagos. In 1996, a group of …