Comments
http://shennessy11.blogspot.com/2011/04/paper-reading-23.html
http://stuartjchi.blogspot.com/2011/04/paper-reading-24-personalized-reading.html
http://shennessy11.blogspot.com/2011/04/paper-reading-23.html
http://stuartjchi.blogspot.com/2011/04/paper-reading-24-personalized-reading.html
Reference Information
Title:Personalized Reading Support for Second-Language Web Documents by Collective Intelligence
Author: Yo Ehara, Nobuyuki Shimizu, Takashi Ninomiya, Hiroshi Nakagawa
Publisher: IUI '10, February 7-10, 2010 Hong Kong
Summary
The researchers in this article developed a way for Web documents to be written in the second languages of users. Today, many users browse Web pages using English word glossing systems, that is, when someone hovers over an unknown word a pop-up window comes up with the definition.
This can be useful but does not take advantage of a user's existing vocabulary. Valuable data is wasted instead of being accumulated. The researchers harnessed collective intelligence by utilizing the accumulated word click logs for many users. Their system automatically predicts unfamiliar words and glosses them with their meaning in advance. If prediction succeeds, the user does not need to consult the dictionary. If it fails, the user can correct the prediction. The predictions are personalized and estimated using a state-of-the-art language testing model that is trained in practical response time with only a small sacrifice of prediction accuracy.
Conclusion
There were a lot of complex equations that I didn't really understand. I'm still not entirely sure how this system actually works. I guess it's a good idea but it's something that I would never use. I also haven't used any English word glossing systems, mostly because I hate pop-overs so much.
Summary
This can be useful but does not take advantage of a user's existing vocabulary. Valuable data is wasted instead of being accumulated. The researchers harnessed collective intelligence by utilizing the accumulated word click logs for many users. Their system automatically predicts unfamiliar words and glosses them with their meaning in advance. If prediction succeeds, the user does not need to consult the dictionary. If it fails, the user can correct the prediction. The predictions are personalized and estimated using a state-of-the-art language testing model that is trained in practical response time with only a small sacrifice of prediction accuracy.
Conclusion
There were a lot of complex equations that I didn't really understand. I'm still not entirely sure how this system actually works. I guess it's a good idea but it's something that I would never use. I also haven't used any English word glossing systems, mostly because I hate pop-overs so much.
awesome paper, might have to read it myself. great analysis
ReplyDelete