AN IMPROVED DEEP LOGANALYSIS METHOD BASED ON DATA RECONSTRUCTION

There are two main difficulties to analyze website user information demand by traditional log analysis method. First, it is difficult to associate the retrieve request with a particular user accurately, so it is unable to reflect the relationship between user characteristics and user behavior. Secondly, it is difficult to extract a complete information interaction process accurately. This paper presents an improved deep log analysis method, whose core idea is to reconstruct the log data by using association relation. The specific measures include behavior association, glossary association and IP association. Experiments have been made to prove that the reconstructed log data is more effective for mining user information demand. The method can solve the defects of the traditional log analysis method, and obtain the good experiment result.