In the Artificial Intelligence community, abstraction is typically viewed as a mechanism for attacking intractable search problems. For example, the performance of a planning algorithm may be improved by using a hierarchy of abstraction spaces. Search algorithms use such abstraction techniques in order to make search control decisions.

Have the two communities missed an opportunity for cross fertilization? For example, do abstraction techniques developed in SE have any application to AI? Do AI techniques for automatically generating abstractions have any application to SE?

• A4 landscape .pdf (80 Ko) for viewing with Acrobate Reader and compressed PostScript .ps.gz (72 Ko) for viewing with Ghostview or printing;
• 4 slides per A4 landscape sheet for printing in color .ps.gz (64 Ko).

\bibitem{Cousot00-SARA-Panel}
P{.} Cousot.
\newblock Contribution to the panel on ``Abstractions in AI and software
  engineering''.
\newblock Fourth International Symposium on Abstraction, Reformulations and
  Approximation, SARA'2000, Horseshoe Bay, Texas, United States, 26--29
  July 2000.

@unpublished{Cousot00-SARA-Panel,
   author =  {Cousot, P{.}},
   title =   {Contribution to the Panel on ``{A}bstractions in {AI} and Software 
              Engineering''},
   note =    {Fourth International Symposium on Abstraction, Reformulations and 
              Approximation, SARA'2000, Horseshoe Bay, Texas, United States, 
              26--29 July 2000},
}