秘书问题研究综述：何时停止搜索信息？  

Review on Secretary Problem Study:When to Stop Searching Information?

作　　者：刘庆顺[1,2] 王渊[1,3] 王刊良[1] 

机构地区：[1]西安交通大学管理学院,陕西西安710049 [2]河北经贸大学公共管理学院,河北石家庄050061 [3]西安工程大学管理学院,陕西西安710049

出　　处：《数理统计与管理》2009年第1期69-81,共13页Journal of Applied Statistics and Management

基　　金：国家自然科学基金优秀创新群体项目(编号：70121001);河北省哲学社会科学基金项目(编号：HB08BSH002)的资助。

摘　　要：秘书问题是一类序贯观察与选择问题,描述了一种动态的信息搜索与决策过程,其问题实质是决定何时停止观察选项、而不是哪一个选项被选择.已有研究成果从解决该问题的策略方法角度,可以分为最优解策略与启发式策略,二者的差异主要体现在理论依据与研究方法上.最优解策略基于决策者完全理性假设,运用数学模型论证了解决该问题的最优决策行为。但许多实证研究发现,人们往往并没有遵循最优决策行为。相比较最优解策略而言,人们通常停止搜索信息太早或者说搜索量太少。这种基于决策者有限理性假设的描述性研究,在解释人们最优选择行为偏离的基础上,提出了解决秘书问题的一些启发式策略.最后,本文通过对已有研究成果的梳理与分析,提出了进一步研究的问题与方向。The Secretary Problem （SP） is a class of decision task on sequential observation and selection in which describes the dynamic decision process with searching information stage by stage, and the essential decision is when to stop searching newer information and not which alternative to be chosen. In the paper, we label optimal policy and heuristic rule to many previous SP researches by the differences of study method and theory foundation. Based on hypothesis of decision-maker＇s complete rationality, the optimal policies proved optimal solutions to SP by mathematics model, but many experiential researches find that people would not conform to the optimal policies to solve SP and often stop searching earlier comparing with the optimal threshold. The descriptive researches based on bounded rationality hypothesis explain the decision behavior of deviation from optimal policy and propose heuristic rules. Finally, we suggest future study problems and perspectives about SP on the ground of reviews and analyses.

关 键 词：秘书问题 决策行为 截止阀法则 

分 类 号：C934[经济管理—管理学] O212[理学—概率论与数理统计]