Kevin Casey
Joe Cabrera
Reference Info:
Mixture Model based Label Association Techniques for Web Accessibility
Muhammad Asiful Islam, Yevgen Borodin, I. V. Ramakrishnan
UIST '10 Proceedings of the 23nd annual ACM symposium on User interface software and technology
Summary:
This paper discusses Internet access for the blind. Currently, the model used is audio based browsing. This is done by assigning text labels with relevant information on a page. This is difficult when dealing with pages with no labels or incorrect labels assigned to them in the HTML. The paper addresses this issue by introducing their Finite Mixture Model algorithm. This associates text with nearby elements. Anything not having a label is assigned a label from a database. An algorithm calculates the likeliest label to be assigned.
Discussion:
I think this sounds like a good idea. Never experiencing this type of technology, I am not sure the issues faced by users needing this. Also, I am not sure how effective just reading off a page is. I feel this is a solution to fix a broken solution, which should be reworked anyway.
I read this paper as well, and am quite curious as to how you would propose to recast this problem, given the limitations of the intended user base. Dynamically-generated Braille? Direct brain-to-computer link à la The Matrix? Personal human web-browsing assistants for all of the blind? ;)
ReplyDelete