Future work will focus on evaluation and exploration of improved techniques for link information presentation, including the value of layered abstractions of Web behaviour. The question of how much users need to know about Web behaviour is a priority for further investigation. The evidence from our user studies confirms that users will generate explanations of Web behaviour and will act upon them. Their reasoning and the analogies they draw upon, however, range from the simple to the sophisticated. So, it is probable that novice users will need different kinds of abstractions from those for more experienced users. Our aim in developing the link lens is to provide support for all kinds of users. An important part of our strategy for achieving this is to recognise that in the hands of a less experienced user, the link lens may serve as a tool for learning about how the Web behaves.
As part of a broader programme of work for improving the value of the Web, we have recently begun to investigate the use of collaborative filtering -- document usage and user ratings -- as additional document quality and relevance information [12,22,26]. As HTTP protocol enhancements become available, we will also examine how they may be applied in this context.
Finally, any mechanism for improving Web users' capacity to find relevant information must ultimately rely upon the willingness of content providers to recognise users' needs. In particular, the need for content providers to include meta-data is paramount. The link lens is designed so that the user will find it easy to distinguish between content providers who are conscientious in this regard and those who are not. In this way, we believe that putting tools like the link lens in the hands of users can provide an incentive for content providers to formulate and adhere to ``good authoring practice''.