An evolutionary algorithm based on compressed representation for computing weak structural balance in large-scale signed networks

Abstract

With the integration of sentimental information, signed networks have a wide range of applications. The calculation of the degree of weak unbalance, which reflects the tension between positive and negative relations, is an NP-hard problem. In this paper, an evolutionary algorithm EAWSB for computing the degree of weak unbalance is proposed, where an indirect individual representation based on compression is designed to reduce the space complexity of the algorithm. In addition, a rotation operator is proposed to increase the population diversity. Experimental results show the effectiveness and efficacy of EAWSB. A thorough comparison show that EAWSB outperforms or is comparable to other state-of-the-art algorithms.