The amount of unstructured data stored in databases around the world severely exceeds the amount of structured data, which can be handled fairly efficiently with current data mining techniques. The utilization and exploitation of unstructured or semi-structured data and its near real-time applications are one of the most important challenges of data mining in the next decade. In this Master’s Thesis I tackle with development of data mining techniques for unstructured data through the motivation of an actual application. My goal is to create a personal web browsing and searching assistant, which creates and inserts live content recommendations in visited web pages, using data mining techniques and on-line analysis of unstructured data. The solution designed, implemented and tested in this Thesis utilizes unstructured data on the user’s visited pages, browsing habits, social connections and interests. With its recommendations, the system aims to accelerate information retrieval, present interesting but unseen content, and make the browsing more fun for the user.