中文摘要：

关键字拍卖竞价策略的制定问题是一个动态多阶段连续拍卖问题.针对拍卖过程中的竞价对手策略空间等信息不完全等问题,文中提出了一种基于优化算法的竞价策略制定Agent模型,并给出该模型的实现方法.该Agent模型将竞价的策略制定问题抽象成为一个多选择背包问题.采用差分进化算法解决多选择背包问题的约束特性.仿真实验表明,该Agent模型的预测模块和优化模块算法选用适当;将该模型应用于交易智能体广告拍卖竞赛中进行比较实验,结果证明该Agent模型具有较高收益性和稳定性等特点.

英文摘要：

Keywords auction theory is an emerging interdisciplinary frontier, which crosses economics, statistics, management sci- ence, and computational science disciplines. For the formulation of keyword auction problem, a biding Agent（ OBS Agent） which is based on the optimization algorithm is presented. And the implementation method of this model is given. The Ad auction problem is abstracted to a combinatorial optimization problem by the OBS Agent, and resolved by optimal algorithm. The results of experiments show that the prediction algorithm and optimization algorithm of the model is effective. The outcome of this model running in the TAC TAC-AA （ trading agent - advertising auction） platform shows that OBS agent has higher income and strong stability.