View article

The paper outlines a framework for automated categorisation of web pages to protect against inappropriate content. The paper contains the framework overview, analysis of state-of-the-art, description of the developed prototype and its evaluation based on series of experiments. Several sources are used for the categorisation, namely text, HTML tags and URL addresses. During the categorisation, this data and other information are analysed using machine learning and data mining methods. Finally, the evaluation of the categorisation quality is performed. The categorisation system developed as a result of this work are planned to be partially implemented in F-Secure Corporation in mass production systems performing analysis of web content.