Personal RECOmmender (PRECO) system is aimed at providing user with easy access to relevant resources located on the Internet. The system is based on semi-automated monitoring of user's activities and behaviour while the user is viewing or reading various resources. Currently only resources in PDF format are supported.

The rationale behind the system is in helping academics and professinals to navigate between the myriad of resources and eliminate the time needed for finding those that might be of interest to the user. The system combines the principles of content filtering and collaborative filtering to identify the potentialy interesting resources.

The system consists of several parts. The one that is interesting for the user is the user agent. The user agent part is responsible for monitoring the user activities while browsing the resources and recording the user's specified level of interestingness. The user agent is incorporotated directly into web browser (currently only MSIE is supported) in the form of Explorer bands and bars so that the interference and overheads are as little as possible. Other parts reside on the server are available as webservices.