The distribution for DAMLOntos, shown in Fig. 2 , is particularly interesting. It seems that there are two areas with different g, or at least a differentiated one on the right. The first one goes from degree 1 to degree 4 with g = -0.43. The second one goes from degree 4 to the maximum degree, i.e. 6380, with g = -3.17. The second one is clear because when it is considered on its own, the error coefficient of the linear regression is only about e_% = 0.75 while for the whole distribution it is e_% = 4.25.

Therefore, we can deduce, as in the BA model, that preferential attachment is taking place in the more connected part of the Semantic Web, those nodes with degree greater than 3. On the other hand, it can be seen that nodes with degree smaller than 4 are basically anonymous resources and literals so it can be said that preferential attachment is taking place in the Semantic Web as a whole.