The debate about which similarity measure one should use for thenormalization in the case of Author Co-citation Analysis (ACA) is furthercomplicated when one distinguishes between the symmetrical co-citation--or,more generally, co-occurrence--matrix and the underlying asymmetricalcitation--occurrence--matrix. In the Web environment, the approach ofretrieving original citation data is often not feasible. In that case, oneshould use the Jaccard index, but preferentially after adding the number oftotal citations (occurrences) on the main diagonal. Unlike Salton's cosine andthe Pearson correlation, the Jaccard index abstracts from the shape of thedistributions and focuses only on the intersection and the sum of the two sets.Since the correlations in the co-occurrence matrix may partially be spurious,this property of the Jaccard index can be considered as an advantage in thiscase.
展开▼
