In this paper we propose a fully unsupervised approach for product aspect discovery in on-line consumer reviews. We apply a two-step hierarchical clustering process in which we first cluster words representing aspects based on the semantic similarity of their contexts and then on the similarity of the hypernyms of the cluster members. Our approach also includes a method for assigning class labels to each of the clusters. We evaluated our methods on large datasets of restaurant and camera reviews and found that the two-step clustering process performed better than a single-step clustering process at identifying aspects and words refering to aspects. Finally, we compare our method to a state-of-the-art topic modelling approach by Titov and McDonald, and demonstrate better results on both datasets.