Online access detection method and device

The invention provides a detection method and a detection device for online access. The detection method comprises the following steps: determining first access data when a user currently accesses a data platform; using a deep learning model to extract first feature data for the first access data, extracting second feature data for the second access data, and identifying the first feature data andthe second feature data to determine whether the current access of the user is abnormal access or not, wherein the second access data is generated when the user accesses the data platform last time.According to the technical scheme, the motivation and intention of current access of the user can be fully mined, and then the accuracy and recall rate of abnormal access behavior detection can be improved. 本发明提供了一种在线访问的检测方法及检测装置，该检测方法包括：确定用户当前访问数据平台时的第一访问数据；利用深度学习模型，针对第一访问数据提取第一特征数据，针对第二访问数据提取第二特征数据，并对第一特征数据和第二特征数据进行识别，以确定用户的当前访问是否是异常访问，其中，第二访问数据是在用户上一次访问数据平台时生成的。本发明的技术方案能够充分挖掘用户当前访问的动机和意图，进而能够提高异常访问行为检测的准确率和召回率。