Query Distribution

[Russ Greiner]

Dear Colleagues,

There are now a number of deployed systems that use belief nets (aka bayesian
nets, probability nets, ...) to answer queries -- ie, to compute the posterior
probability of some variable(s), based on some specified set of evidence.  It
would be very useful to know the actual distribution of queries posed to such
real-world systems; eg, how often the user asks
   "What is the probability of cancer, given Fever=T and Age>42 ?",
vs 
   "What is the probability of cancer, given Fever=F, lump=F and Gender=M ?"
vs
   "What is the prior probability of hepatitis ?"
etc etc etc.
We could then use this "query distribution" to evaluate our learning
algorithms, by computing (perhaps) the 

   *average (sum-squared) accuracy* 

of the belief net it returns, where the "average" is wrt this
real-world distribution (cf, [Greiner/Grove/Schuurmans, "Learning 
Bayesian Nets that Perform Well", UAI-97]).

We are therefore looking for some real-world *query distributions*.

Please let me know if you can provide this information -- perhaps in the 
form of the set of queries actually posed to a real system, or a set of
session transcripts or log files, of a system's interations with its users, 
or ... 

To avoid confusion, note that this QUERY DISTRIBUTION cannot necessarily be
inferred from the given belief net B, as the query distribution might be
completely unrelated to the "NATURAL DISTRIBUTION" of events (encoded by B).
Eg, we may ask many queries about low probability events --- the probability 
of the QUERY
   "What is the probability of cancer?"
may be very high, even though the actual probability of 
    Cancer
is very low.

Thank you.

| Russell Greiner		Phone:    (403) 492-5461            |
| Dep't of Computing Science	FAX:      (403) 492-1071            |
| University of Alberta		Email:	  greiner at cs.ualberta.ca    |
| Edmonton, AB T6G 2H1 Canada	http://www.cs.ualberta.ca/~greiner/ |