Aspect identification in user generated texts by supervised text classification might suffer degradation in performance when changing to other domains than the one used for training. For referring to aspects such as quality, ...