An Interactive Self-Learning Multi-Agent Genetic Algorithm for the Travel Itinerary Planning Problem
Combining the multi-agent system with the interactive genetic algorithm, an interactive self-learning multi-agent genetic algorithm for the travel itinerary planning problem is proposed in this paper. The algorithm makes agents (individuals) of the population fixed on a circle two-dimension lattice evolve and compete in order to search the satisfactory travel itinerary. A user only needs to select and mark two agents (the current best agent and the second best agent) in every generation, not to evaluate all agents. And the current best agent can carry on a self-learning to make it improved locally with a hill-climbing search. The algorithm reduces the number of total times of the user’s evaluation, and contributes to relieve the user fatigue. The experiment shows that the algorithm is feasible, effective and efficient in the travel itinerary planning problem.
Interactive genetic algorithm Multi-agent Travel itinerary planning User fatigue
Changyong LIANG Qing LU Yongqing HUANG
Institute of Computer Network,Hefei University of Technology,Hefei 230009,China
国际会议
北京
英文
2007-05-30（万方平台首次上网日期，不代表论文的发表时间）