Web servers store every event in the form of logs, which contain the retrieved URL, client IP address, access time, HTTP status code, etc. There are several useful methods to analyze these log files, which are mainly based on sequential text mining techniques for applications like prefetching or driving critical information about system’s security. Finding ways to convert web server log files into graph structures may open new horizons in complex network analysis for investigation, comparison, and prediction. Since visibility graph has various successful usages for analyzing different time series data, web server log files as a kind of time-series data have the potential to be converted into visibility graph. In this research, we propose a novel method to convert web server log files into horizontal visibility graphs. Afterward, we demonstrate the result of the method on two popular datasets, NASA and Online Judge web server log files, and perform exploratory and visibility graph analysis techniques like centrality measures computation and community detection to show the promising future for the research. Moreover, we introduce a novel algorithm for a common application in web server log file analysis, web prefetching, based on a modified version of link prediction on the extracted visibility graph, and evaluate it based on AUC assessment and propose the next page to prefetch in each dataset. Finally, we propose several choices to extend the research in case of technical and practical aspects.
