Web Nav : a context based navigation assistant for the web

Abstract

Today's Web challenge is not simply of just finding information, but also refinding information. Studies in the area suggest that page revisitation accounts for around 58% to 81% of all page visits. Studies also suggest that current browsers are not well as well equipped to cater for this revisitation as they could be. We take the approach of trying to improve navigation on the Web in terms of both revisitation and locating new information, hence combining elements of a history mechanism and a page recommender system. We present an underlying framework built on the concept of paths through hyperspace, and their persistence in entirety. Moreover we group these paths in terms of their context to form context based subspaces, which are used both as a history mechanism and as a basis from which queries are generated to locate new information relevant to the context. We envision a tool that caters for resuming of a context by simply jumping back into any point in the corresponding path, regardless of session boundaries. Our user interface is inspired from systems such as WebView and WebNet, which seek to improve the visual element of history by going beyond the norm by presenting history through maps and views. We also implement a cover flow view of page thumbnails inspired from popular implementations such as in Apple iTunes. Initial findings through a preliminary evaluation suggest the presence of a significant obstacle for users to shift from the Back/Forward stack based paradigm to a tree-based paradigm centred around the clustering of paths. Evaluation data shows that mixed but mostly positive reactions to the project and the features offered. While revisions are naturally necessary, as well as a more comprehensive evaluation to examine inference processing carefully, our initial conclusions are that the paradigm is promising, with the majority of evaluators indicating an interest in shifting to this new paradigm in the future. We also surmise that such a framework for saving path information might be useful for further path based initiatives in the areas of Web movement archiving and virtual tours.