A Novel Semantic User Operation Restoration from Massive Web URL Log

In large distributed management information system, user operation logs are often implicit but cannot directly reflect the business semantics of user operations. In order to meet the needs of operation and maintenance personnel to carry out work such as security detection and exception analysis based on the business significance, this paper designs a business semantic restoration method for log URL access sequences based on fasttext algorithm. In our method, the original log data is processed to form a structured sequential log record with users as the unit. The log records of one single user will be mapped into a URL access sequence, and this sequence is divided into several business text segments which are composed of URLs and are associated with context semantics. Finally, through the supervised text classification algorithm fasttext, we will get a model that can recognize the business attributes of the text segment of the user URL sequence, in other words, realize the business semantic restoration of the user operation log sequence, which provides data support for the system operation and maintenance personnel.