From: rjones@colby.edu   
      
   Sonia wrote:   
   > Hello guys.  I am new to AI and I'm researching these 2 questions, for my   
   > own use, it's not an assignment, that is why I need your help.  Since this   
   > will not be graded by anyone, I just want to make sure that I am   
   > understanding the problems correctly.   
   >   
   > 1. What role does domain-specific knowledge play in AI ? must   
   > domain-specific knowledge be used for an approach to be considered an AI   
   > problem? If domain specific knowledge is used, list   
   > 3 major AI techniques and indicate where in each technique the domain   
   > knowledge is represented in and in what form.   
   >   
   > 2. an important aspect of some AI domains is the use of knowledge from human   
   > domain experts. discuss 2 domains in which such knowledge is used and   
   > identify ways in which knowledge is acquired and used in both domains.   
   >   
   > for problem #1, from what I've read it seems that the domain-specific   
   > knowledge must be used in order for the approach to be considered an AI   
   > problem.  I cannot think of any AI approach that does not use   
   > domain-specific knowledge.  Do you ?   
   > Now if that's the case, what 3 techniques use domain-specific knowledge.   
   > Would an example of that be decision trees, expert systems, neural nets,   
   > genetic algorithms, ??and possibly something else ?  and as far as   
   > representation goes, they are either represented as propositional, predicate   
   > logic, semantic nets and frames.  Is that correct ? I am not sure about this   
   > one .. please let me know about this one   
      
   I guess it would be difficult to imagine a deployed AI application that   
   didn't incorporate domain knowledge.  But there's lot's of theoretical   
   work that is not tied to a particular domain, but is still called "AI".   
     As a simple example, the A* search algorithm is a "weak method",   
   meaning it has no particular domain knowledge built into it, and you can   
   make it work for whatever domain you want to.  But to actually use it   
   for a particular problem, you have to add some domain knowledge, in the   
   form of the state and goal representations, the operators for moving   
   from state to state, and the heuristic for evaluating states.   
      
   >   
   > for problem #2, would those be expert systems, decision trees, neural nets ?   
   > are those examples of such, and are there any others ?   
      
   Maybe there's a terminology problem here, but I wouldn't call those   
   "domains".  In my experience, in AI a "domain" refers to a particular   
   task, where what you have listed are "approaches".  For example, playing   
   chess, solving probability problems, or flying and airplane could each   
   be considered a "domain", and each comes with their own requirements for   
   specific "domain knowledge".   
      
   [ comp.ai is moderated.  To submit, just post and be patient, or if ]   
   [ that fails mail your article to , and ]   
   [ ask your news administrator to fix the problems with your system. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)