Nowadays netizens embark on a prevalent lifestyle to actively voice out their opinions online that includes both forums and social networks (Web 2.0). Their opinions which initially are intended for their groups of friends propagate to attentions of many. This pond of opinions in the forms of forum posts, messages written on micro-blogs, Twitter and Facebook, constitute to online opinions that represent a community of online users. The messages though might seem to be trivial when each of them is viewed singularly; the converged sum of them serves as a potentially useful source of feedbacks to the current affairs after analysis. A local government, for instance, may be interested to know the response of the citizens after a new policy is announced, from their voices collected from the Internet. However, such online messages are unstructured in nature, their contexts vary greatly, and that poses a tremendous difficulty in correctly interpreting them. In this paper we propose an innovative analytical model that evaluates such messages by representing them in different moods. The model comprises of several data analytics such as emotion classification by text mining and hierarchical visualization that reflects public moods over a large repository of online comments.