
Research Topic: Dusty web pages to support social navigation
Suggested By: Jon Dron
Last Update: May 16th, 2025
DustThe ideaWebsites are presented with a thin layer of 'dust' on them that slightly veils the content with a washed-out haze (actually a semi-transparent image generated dynamically by the application). When users click on the page or select a hyperlink using the keyboard, the click/keypress is passed through to the page below but, at the same time, the application stores that information in a database and clears some of the dust from the image it presents for that page. Dust continues to fall regularly whether or not the page is clicked so, if no one else clicks on it, it becomes cloudy again. The point of the system is to use collective wisdom to identify points of interest and links that are important to the crowd. An interesting augmentation would allow site owners to explicitly rub dust away from their pages - they would only be likely to do this if the pages were actively maintained, so it would be simple to see pages that are not cared for. The undifferentiated crowd can be a bit of a mob and results may be less than relevant to an individual seeking navigation support. A further useful modification would therefore be to offer different views of the dust to different communities. For example, if a given page were used by two class groups (from different courses in a university, say) then each group would only see the dust rubbed away by their own group. Taking this still further, a collaborative filter (based perhaps on user profile characteristics or previous behaviour) could be used to weight the clicks so that those most similar to the user would have greater cleaning power than those identified as dissimilar. ImplementationJavaScript used to capture clicks and selection of hrefs (e.g. via keyboard) Application generates semi-transparent PNG (or transparent with small dots) to be overlaid on page. Algorithm for dust accumulation is self-tuning according to regularity of visits to a given page - less visited pages accrue dust more slowly. Two main potential forms of implementation:
Whichever form of implementation is used, there are various research issues to be addressed here, some technical, some more to do with usability, accessibility and behaviour. This is a 'soft' recommender system, whereby users are better able to identify resources that may be useful to them, without affecting their ability to navigate to other resources if they wish (contrast this with a 'hard' recommender such as Amazon's book recommendations, where only a very small filtered subset of recommendations is supplied). There are many issues relating to how to most effectively guide users, whether the recommendations are effective, how such things are perceived in terms of usability, how to prevent mob stupidity, the balance of negative and positive feedback and so on. |