Scope of the thesis is to design and test a new approach for book recommendation portal. The purpose of the project is to examine, how users react to the different way of book recommendation, which is based on presenting small piece of the story from the book itself without confronting user with name of the author, book title or general description. This approach assumes to investigate if accuracy of recommendations in already existing recommendation portals based on the different algorithms and already verified procedures is effective enough to provide user satisfaction. To answer these questions it is planned to do few stages of iterations and user testing. The assumptions will be also well-funded by the literature study and problem analysis. Part of the project will be also creating of concept of such a recommendation portal. Motivation for this project is to build an innovative book recommendation system based on the random book browsing, like people used to do in libraries and bookshops.