Except for a handful of "mobile" Web sites, the Web is designed for browsing using personal computers with large screens capable of fully rendering the content of most Web pages. Browsing with handhelds, such as small-screen PDA's or cell phones, usually involves a lot of horizontal and vertical scrolling, thus making Web browsing time-consuming and strenuous. At the same time, one is often only interested in a fragment of a Web page, which again may not fit on the limited-size screens of mobile devices, requiring more scrolling in both dimensions. In this paper, we address the problem of browsing fatigue during mobile Web access using geometric segmentation of Web pages and the notion of context. Our prototype system, CMo, reduces information overload by allowing its users to see the most relevant fragment of the page and then navigate between other fragments if necessary. On following a link, CMo captures the context of the link, employing a simple topic-boundary detection technique; then, it uses the context to identify relevant information in the next page with the help of a Support Vector Machine, a statistical machine-learning model. Our experiments show that the use of context can potentially save browsing time and improve the mobile browsing experience.
