It is well known that supervised text classification methods need to learn from many labeled examples to achieve a high accuracy. However, in a real context, sufficient labeled examples are not always available. In this paper we demonstrate that a way to obtain a high accuracy, when the number of labeled examples is low, is to consider structured features instead of list of weighted words as observed features. The proposed vector of features considers a hierarchical structure, named a mixed Graph of Terms, composed of a directed and an undirected sub-graph of words, that can be automatically constructed from a set of documents through the probabilistic Topic Model.