This page is a demonstration of a text classification approach to estimating the reading difficulty of text.
Reading difficulty estimates for very short texts (e.g., 1-10 words) tend to be less accurate than estimates for longer texts. Accuracy generally improves with text length. The software will provide estimates for texts of any length, but a minimum length of 20 words is recommended.
Type or paste your text into the box below and press "Submit" to obtain an estimate of the difficulty of your text.
For more information, see:
K. Collins-Thompson and J. Callan. (2005.) "Predicting reading difficulty with statistical reading models". Journal of the American Society for Information Science and Technology, 56(13) (pp. 1448-1462).
K. Collins-Thompson and J. Callan. (2004.) "A language modeling approach to predicting reading difficulty." In Proceedings of the HLT/NAACL 2004 Conference. Boston.
The Reap project.
|
|
This research was sponsored by the National Science Foundation and the U.S. Department of Education's Institute for Education Sciences. Any opinions, findings, conclusions or recommendations expressed on this Web site are those of the author(s), and do not necessarily reflect those of the sponsors. | |
Updated on April 28, 2006.
http://www.cs.cmu.edu/~callan/