Difference: AssistKidsToBrowseWeb (1 vs. 2)

Revision 22009-10-15 - NaushadUzZaman

Line: 1 to 1
 
META TOPICPARENT name="NaushadUzZaman"

Assist Less Educated People and Children

Revision 12009-09-22 - NaushadUzZaman

Line: 1 to 1
Added:
>
>
META TOPICPARENT name="NaushadUzZaman"

Assist Less Educated People and Children

Application:

  • Help Less Educated people to browse web by suggesting images for keywords
  • Help Children to browse and read
  • Help people learning a new language (e.g. English) [an assisting application for novice English as second language students]
Approaching the problem:
  • Breadth First:
    • Start with English [keyword extraction problem (given a document extract the keywords) then showing images for keywords]
    • Extend with Bangla [language with some resources and i know the language]
    • Try to generalize with other languages [main challenge: not all languages have resources as English]
  • Depth First:
    • Only with English
    • Go deeper in English and try to give user a graphical view of how events are inter-connected

Meeting with Jeff (09/09/09):

- Focus first on children in developed countries than less educated people in developing country, because the task would be harder to get them use it (definitely our very next step would be to reach less educated people as well)

Areas to explore:
  • AI - NLP (text to keyword to picture, or text to picture)
  • HCI (Accessibility)
  • Education and Brain & Cognitive Science (How to help kids read something with pictures)
Another new application: Approaching the problem:
  • Previously considering one picture for a document -> consider picture for each paragraph
  • When we can understand the document then connect the pictures of a paragraph, i.e. each group of picture will link to a paragraph
  • When user clicks the image then it will start reading that particular paragraph with highlighting the text its reading

Meeting with James (09/09/21) - ideas discussed in meeting with Jeff

  • text to image is very hard
  • for each paragraph generate an image
  • try to extract the agents and different entities in the paragraph and connect between entities
  • e.g. obama says something about health-care then extract "obama" and "health care" and show obama is saying something about health-care
  • one particular challenge this approach tries to handle is, getting an image that explains a paragraph is very hard - so this will at least try to get the concepts and connect the concepts to represent something useful
  • as we discussed before, clicking this image will go to the paragraph and start reading
References to check:
Entity Detection and Tracking
http://www.cs.utah.edu/~hal/docs/daume06thesis.pdf

Florian et al., 2004;
Florian, R., H. Hassan, A. Ittycheriah, H. Jing, N. Kambhatla, X. Luo, N. Nicolov, and S. Roukos. 2004. A statistical model for multilingual entity detection and tracking. In Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics and Human Language Technology (NAACL/HLT).


 
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2017 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding URCS? Send feedback