“…We select topics with the following value ranges: [2,7], [5,10], [5,100], [10,15], [10,50], [15,20], and [50, 100], where the first value corresponds to the minimum number of documents and the second value corresponds to the maximum number of documents. With these values, we explored different cluster granularities, which, in turn, allowed us to observe the specificity of the aspects; those produced in (sub)clusters with [2,7] documents are more specific, while those produced in (sub)clusters with [50, 100] documents are more generic. By experimentation, the granularity [2,7] provided the best results, as it contains more specific descriptions about the items.…”