Integrating Linguistic Resources in an Uniform Way for Text Classification Tasks