Recommender systems play an important role in several web applications. Their essential goal is to pick some items from the continuously growing amount of content and recommend them to users who are likely to be interested in them. One of the main types of recommendation is implicit recommendation, in which we do not assume that the users differentiate between items in an explicit way (i.e. in the form of ratings) but instead, we have some implicit information about the interaction between users and items. A special case of this is „item-to-item” implicit recommendation, where the related activity of the user is only known within a limited period of time. This can occur because of the fact that most users are unlikely to sign in to a website unless it is necessary.
The goal of my thesis is to create a content-based implicit recommender system. For this, I use a dataset containing movies and some ratings by users.
In the first section of my thesis, I present the theoretical background required to reach my goal, the main types of recommender systems, some possible ways of recommendation and how they are used in some of the most popular web applications.
The second part of my thesis focuses on the implementation. In this part, I present the chosen technologies, the implementation itself and the process and results of testing the application, together with further possible ways of improvement.