An ever-increasing number of operations is performed nowadays with the help of computer systems. They are everywhere around us and each of them produces a huge quantity of log files for each operation. In this context, the aim of this project is to study and create an architecture able to read and categorize the logs produced by computer systems based on logs content. The outcomes of this project form the foundation of a subsequent task of anomaly detection, which is highly requested since lets companies to detect malicious attacks or failures and by which the focus is moved towards the most relevant events.