Abstract
Genetic algorithm (GA) has been widely used in combinatorial optimization problems, but direct application of GA to grouping problems often turns to fail. This difficulty arises from the large amount of constraints that are intrinsic to the problems of this category. Following this observation, this paper adopts an alternative encoding based on the “grouping genetic algorithm” (GGA) proposed by Falkenauer. The whole algorithm consists of two stages : session construction and session allocation. Keywords and category keywords are used for grouping. The experiments using real data will validate the usefulness of the proposed method.
