Agent-based adaptive travel planning system in peak seasons

Abstract

With the wide spread of Internet, intelligent systems to support travel planning have been progressed during last decade. The current on-line travel support systems, however, have limits in their capabilities to adapt to the changes in travel plans. When the preferred travel plan is not available, especially during peak seasons such as vacations, holidays, and weekends, a traveler needs to re-plan by compromising his or her preferences, which requires time consuming and cumbersome efforts. This dissertation introduces an adaptive travel planning system with intelligent agent that can adapt to the dynamics of travel plan. This system consists of an adaptive travel planning model, which is served as knowledge of agents to find alternatives and choose the best one, and a collaboration mechanism between the agents. To adapt to the dynamics of travel plans, the adaptive travel planning model has the following features: finding alternatives by constraint relaxation, formulating a decision problem to select the current best alternative, and recommending the best alternative at each decision time. This model makes travel agents provide more acceptable travel schedules to travelers while reducing the risk of having no ticket. In particular, an adaptive collaboration mechanism between the agents is designed to handle the process of adaptive travel planning. This mechanism makes each autonomous agent adapt to changes in the availability of travel plans while enhancing the efficiency. A prototype is implemented on the standard agent specification of FIPA which can be used for the automation of dynamic and complex travel planning. The usefulness of the system is illustrated using a simulation analysis and a specific scenario of travel planning.