信息检索决策中的学习规则量化分析——强化模型拟合实验观测

Quantitative Study on the Characteristics of Information Users＇ Searching Decision-making Rules Based an Experiment About Reinforcement Learning Models

使用学习模型定量分析信息用户检索行为特征,对于扩展传统用户研究的方法、促进学科间融合具有重要意义。本文首先分析了信息用户检索决策中的强化学习特征,并选择了强化学习模型中的Bush-Mosteller模型和Borgers-Sarin模型对用户的检索决策行为形成过程进行模拟。模拟结果表明Borgers-Sarin模型成功捕捉了用户检索行为的形成过程,说明用户学习规则特征表现为将预期与反馈结果进行比较;模型的参数表明用户行为的黏滞性、惯性特征明显。

It is of great meaning to analyze information users＇ behavior by using learning models for expanding traditional search methods and promoting amalgamation among different subjects.Based on analyze information users＇ reinforcement learning characters,this paper choose two reinforcement learning models,Bush-Mosteller model and Borgers-Sarin model to simulate decision-making behavior.The results show that Borgers-Sarin model successfully catch the characters of information users＇behavior,which is anticipation is important for behavior forming and the parameter of the model show that the inertia of the behavior is obvious.