The conformation of opinion in a society is analyzed in diverse studies, following diverse perspectives, methods and tools. Previous studies are basically descriptive, predictive, and not aiming at modeling the real situation. In this paper, we present a preliminary model representing the process of conformation of public opinion from the perspective of opinion mining, by using the Multilevel Fuzzy Cognitive Map (MFCM), which, in turn, is employed to evaluate the quality of the public opinion. This paper examines the capabilities of such model, for both, knowledge description and knowledge extraction about opinions presented in a certain topic, allowing the assessment of the quality of public opinion. This is very valuable for decision making, as well as for social and political analysis. We give special attention to the influence of the media on public opinion, because of its importance in the actual western society. The design of the MFCM, its concepts and relationships, are presented. The concepts describe the relevant aspects from which public opinion emerges, covering the social, technological and biological domains. The results are encouraging: the model permits to identify the media preferred by the public in order to understand better significant matters, including the high esteem that the new communication media hold, in contrast with the little merit that the traditional communication media enjoys, as well as the confidence and credibility in the information sources (from both, new and old media) diminish in time.