Web Log Mining Based-on Improved Double-Points Crossover Genetic Algorithm

Web log files have become important data source for discoveries of user behaviors. Analyzing web log files is one of the significant research fields of web mining. This paper proposes an improved double-points crossover genetic algorithm for mining user access patterns from web log files. Our work contains three different components. First, we design a coding rule according to pre-processed web log data. Second, a fitness function is presented by analyzing user sessions. Finally, a new genetic algorithm based on double-points crossover genetic algorithm is designed. In comparison with simple genetic algorithm, double-points crossover genetic algorithm demonstrates better convergence than the other, and it is more suitable for web log mining. We conducted an experiment to verify the effectiveness of the proposed algorithm. The results show that the proposed algorithm helps the website to easily gain access patterns.