情境大数据建模及其在用户行为预测中的应用

Modeling contextual big data for user behavior prediction

随着大数据时代的到来,信息系统收集了海量情境信息,如舆情信息、环境信息、经济信息等。这些情景大数据提供丰富的细节信息,更细致地刻画行为背景以辅助用户行为建模。阐述了两种使用表达学习策略建模一般化情境信息的框架,并针对情境大数据中最常见的时序情境建模问题,使用循环神经网络建模时序情境中的序列依赖关系。

In the big data era, information system has to handle a mass of data of contextual information, such as public opinion, environment information and economic status. Embedded with abundant details of user behavior, contextual information plays a significant role in effectively shaping user character and elaborately modeling user behavior. Two frameworks to model general context information through representation learning and a recurrent model for sequential context scenarios were involved.