Analysis of Web Logs: Challenges and Findings

Web logs are an important source of information to describe and understand the traffic of the servers and its characteristics. The analysis of these logs is rather challenging because of the large volume of data and the complex relationships hidden in these data. Our investigation focuses on the analysis of the logs of two Web servers and identifies the main characteristics of their workload and the navigation profiles of crawlers and human users visiting the sites. The classification of these visitors has shown some interesting similarities and differences in term of traffic intensity and its temporal distribution. In general, crawlers tend to re-visit the sites rather often, even though they seldom send bursts of requests to reduce their impact on the servers resources. The other clients are also characterized by periodic patterns that can be effectively represented by few principal components.