Constructing method of time series database for cognition early-warning

The invention relates to a constructing method of a time series database for cognition early-warning, wherein an information dataset is formed through acquiring real-time action information of a userand integrating the real-time action information. The constructing method further comprises the following steps of executing higher-level classification on the information dataset for constructing anexpert knowledge database in a manner of acquiring multiple classification data; executing construction of a behavior predicting classifier based on the classification data for predicting a behavior which will occur at a preset time point in future; performing scoring processing on the predicted behavior, and generating early-warning information when a score reduction magnitude exceeds a preset threshold. 本发明涉及一种用于认知预警的时序数据库的构建方法，通过采集使用者的实时动作信息并对其进行整合以构成信息数据集，所述构建方法还包括如下步骤：按照对所述信息数据集执行向更高层次的分类以获取若干个分类数据的方式构建专家知识数据库；基于所述分类数据而执行行为预测分类器的构建以对未来设定时刻所发生的行为进行预测；针对预测获取的行为进行评分处理，在评分下降超过设定阈值的情况下生成预警信息。