Combining local search and look-ahead for scheduling and constraint satisfaction problems

  • Authors:
  • Andrea Schaerf

  • Affiliations:
  • Dipartimento di Informatica e Sistemistica, Universita di Roma "La Sapienza", Rome, Italy

  • Venue:
  • IJCAI'97 Proceedings of the Fifteenth international joint conference on Artifical intelligence - Volume 2
  • Year:
  • 1997

Quantified Score

Hi-index 0.00

Visualization

Abstract

We propose a solution technique for scheduling and constraint satisfaction problems that combines backtracking-free constructive methods and local search techniques. Our technique incrementally constructs the solution, performing a local search on partial solutions each time the construction reaches a dead-end. Local search on the space of partial solutions is guided by a cost function based on three components: the distance to feasibility of the partial solution, a look-ahead factor, and (for optimization problems) a lower bound of the objective function. In order to improve search effectiveness, we make use of an adaptive relaxation of constraints and an interleaving of different lookahead factors. The new technique has been successfully experimented on two real-life problems: university course scheduling and sport tournament scheduling.