Comparing monolithic and committee-based profiles for politician recommendation.

In a parliamentary setting, citizen could be interested in knowing those Members of Parliament (MPs) who are working in different areas or involved in the resolution of some people's problems. These topics of interest are usually represented by means of a profile. In this paper, the politicians' profiles are built considering the speeches in parliamentary sessions. However, in most of the cases a single profile is not the best alternative to represent MPs' interests because the specific terms related to a given topic are mixed with others, so that the MPs' preferences are diluted. The alternative is to build different subprofiles containing each one the most representative keywords for each topic, creating in this way a richer representation. We present a first approach to build subprofiles based on the MPs' speeches in different committee and plenary sessions, which will be compared, in terms of performance, to monolithic profiles for an MP content-based recommendation task.