中文摘要：

大数据环境下,政府信息公开主体的行为更加容易被计量、预测和分析。文章运用委托代理模型,量化分析上级政府针对单个下级政府、多个下级政府在政府信息公开过程中的互动行为,并探究影响互动方式的主要因素及其所产生的影响。研究表明：下级政府的信息公开比例（信息占有量的影响程度小于信息公开比例）和努力程度是最重要的两项指标;对于上级政府而言,量化下级政府的成本系数、努力程度可以优化政府信息公开行为;对于下级政府,增加自身信息公开的努力程度比降低自身信息公开的成本对自身收益的影响更大。

英文摘要：

Under the big data environment, the behavior of government information disclosure is more easily measured, predicted and analyzed, and then could be quantitative analyzed and optimized. By using the principal-agent model, explore the interactive behavior of higher levels of government to single lower levels of government individual and multiple lower levels of government＇s individuals respectively, and to explore the influence factors and the main effect of interaction. Conclusions： information disclosure proportion of lower levels of govemrnents （The impact of information possession is less than the proportion of information disclosure） and the degree of effort are the two main indicators. In addition, for the higher level government, the cost coefficient and the degree of effort can optimize the government information disclosure behavior; for lower levels of governments, increase the degree of information disclosure efforts have the greater impact on income than reduce the cost of theft own information disclosure.