Concepts and methods of complex networks have been employed to uncoverpatterns in a myriad of complex systems. Unfortunately, the relevance andsignificance of these patterns strongly depends on the reliability of the datasets. In the study of collaboration networks, for instance, unavoidable noisepervading author's collaboration datasets arises when authors share the samename. To address this problem, we derive a hybrid approach based on authors'collaboration patterns and on topological features of collaborative networks.Our results show that the combination of strategies, in most cases, performsbetter than the traditional approach which disregards topological features. Wealso show that the main factor for improving the discriminability of homonymousauthors is the average distance between authors. Finally, we show that it ispossible to predict the weighting associated to each strategy compounding thehybrid system by examining the discrimination obtained from the traditionalanalysis of collaboration patterns. Once the methodology devised here isgeneric, our approach is potentially useful to classify many other networkedsystems governed by complex interactions.
展开▼
