Principal Investigator:


C/ Tajo, s/n, 28670 Villaviciosa de Odón



The system is a custom server MERCURIO News working with a client representation from the latest advances in user modeling. The news server is developed as a Java application that receives customer subscriptions through a website. During the registration process the customer specifies their preferences when receiving news, and with them creates a user model that is used to send you news that may be of interest to the frequency you specify.The news server is also cooperating with a search engine that allows customers to perform specific searches on the news of the day.


A user connects to the server and recorded information in the system. During registration a number of requested personal information (email address, login and password). After building a profile (or model) for each user, which contains the following information: days of the week that you want to receive news, news maximum number of user message and interests. The system applies the user model to the news of the day, using text classification techniques [1, 2]. It calculates the relevance associated with each story and construct a ranking. The most important news are sent to the user. As an additional service, users can search the news of the day.


The MERCURIO system is organized according to the tasks they perform:

  • Register or from the service, or edit the user profile.
  • Receive a personalized message every day selected by the user, containing the most interesting news for him.
  • Search the news of the day.

The user model component manages user profiles and how they are organized. A user is represented by a set of information that characterize it:

  • Information (name, login, password, email address).
  • Information about your preferences (days of the week you want to receive messages, maximum number of news per message, temporarily disabling the service).
  • Information about the user’s interests: (1) sections, (2) general categories, (3) terms.


The MERCURIO system implements a custom shipping news based on a user model that captures the different user interests. With this model and text classification techniques are filtered news relevant to a particular user. The proposed system has all the desirable features for these systems: the use of more information 
sections and terms to describe user’s interests, effectiveness, personalization, ease of use and lack of noise.