In this paper, we described three techniques for bridging the web accessibility divide
between sighted and visually impaired users. First, we described the design
of context-directed browsing approach, which uses Web page partitioning and techniques
from NLP and Machine Learning. Using our system, visually impaired individuals
can potentially imitate the browsing behavior of sighted users, saving their
time on not listening to irrelevant information. Contextual browsing also has implications
for handheld devices with small screens. We implemented context-direct
browsing algorithm to identify and display the most relevant sections of Web pages
on small-screen handhelds [48].