An AHP-TOPSIS Based Framework for the Selection of Node Ranking Techniques in Complex Networks

Kushal Kanwar, Sakshi Kaushal, Harish Kumar

Abstract: Disparate natural and artificial systems are modelled as complex networks to understand their structural properties and dynamics of phenomena occurring on them. Identification of key components (nodes) of a complex network and ranking them has both theoretical and practical applications. The node ranking techniques are compared on three categories of criteria, namely, Differentiation, Accuracy and Computational Efficiency. Having multiple criteria for technique selection and a number of alternative ranking techniques available renders ranking technique selection in the domain of complex networks as Multi-Criteria Decision Making (MCDM) problem. A number of MCDM methods are accessible in the literature, but no treatment is available for the selection of node ranking techniques based on systematic decision making. In this paper, Analytic Hierarchy Process (AHP), followed by Technique for Order Preference by Similarity to Ideal Solution (TOPSIS), are used in tandem to propose a framework that objectively compares and select node ranking techniques for complex networks. The working of the proposed framework is demonstrated with a dataset of complex networks.