Dear Patrick,
you might want to take a look at the paper
@InProceedings{
author = "Giuseppe Carenini and Stefano Monti and Gordon Banks",
title = "An Information-based Bayesian approach to history-taking",
booktitle = "5th Conference of AI in Medicine, Europe",
year = 1995,
pages = "129--138",
address = "Pavia, Italy",
annote = "http://www.cs.cmu.edu/~smonti/HTML/documents/aime95.html"
}
where we use a combination of expected value-of-information and "expected
entropy" to conduct a questionnaire-based evidence gathering process, while
trying to minimize the number of questions asked (the entropy was used in
those cases when the value of information wouldn't help, see paper for
details).
Although it is not applied to adapting a questionnaire, the Pathfinder's
paper by Heckerman et al. is also relevant (they use a decisin theoretic
approach to the sequencing of clinical tests to arrive at the "optimal"
differential diagnosis):
@Article{hechor92,
author = "David Heckerman and Eric J. Horvitz and B.N. Nathwani",
title = "Toward Normative Expert Systems: Part I The Pathfinder
Project",
journal = "Methods of Information in Medicine",
year = "1992",
volume = "31",
number = "2",
pages = "90-105"
}
Hope it helps. Cheers,
-- ste
Patrick Naim wrote:
> Hello,
> I am looking for any work/paper on building adaptive questionnaires
> using a prior bayesian model.
> The typical application would be online credit application. The idea is
> that you decide to accept/reject the application whenever the expected
> score is high/low enough with sufficient probability, based on collected
> answers, regardless of the next answers. So the question is how to chain
> questions so that the questionnaire comes to a yes/no conclusion as
> quick as possible, with good probability of success.
> Thank you in advance.
> Patrick Nam
___________________________________________________________________
Stefano Monti | Voice: (412) 268-3740 Fax: (412) 268-5569
CMU, Robotics Institute | Email: smonti+@cs.cmu.edu
Pittsburgh, PA 15213 | http://www.cs.cmu.edu/~smonti
------- End of Forwarded Message
This archive was generated by hypermail 2b29 : Fri May 19 2000 - 12:22:16 PDT