Data Mixed-Extraction Strategy Based on the Time Characteristics in CDW

There is a great deal of general information and behavior information of customers stored in clickstream data warehouse, so it has lots of data sources. Data extraction technologies such as traditional web server logs and packet sniffer have many inadequacies in the extraction efficiency and accuracy. According to the actual needs of OLAP and DM, this dissertation proposes a hybrid data extraction strategy based on time characteristics in the server layer, which makes up the shortcomings of clickstream data such as data incompleteness by combining clickstream data and external data sources.