مجله بین المللی نوآوری در علوم کامپیوتر و فناوری اطلاعات 2 3 Improving Influence Propagation in Social Networks using Fitness Sharing Algorithm with Dynamic Sharing Radius Improving Influence Propagation in Social Networks using Fitness Sharing Algorithm with Dynamic Sharing Radius 42 51 10.22051/jera.2021.31891.2698 FA Fereshteh Sheikhbahaee University, Isfahan, Iran Navid Sheikhbahaee University, Isfahan, Iran Mehdi Sheikhbahaee University, Isfahan, Iran In social networks, related people are influenced by each other because people share their ideas and views on different issues. If a number of people in a network adopt a particular behavior or belief, this behavior or belief is spread in the network due to the social relationships within the network. This phenomenon is called Influence Propagation. One of the most important issues in influence propagation optimization is the issue of influence maximization. In influence maximization, the goal is to find k subset of members of the social network so that by activating them, under an information diffusion model, the largest number of network members will be affected by information. The purpose of this study is to provide a solution to find the most influential people using the fitness sharing algorithm with dynamic sharing radius and under the Linear Threshold Model. The proposed solution is one of the meta-heuristic solutions in which the genetic algorithm is used. The proposed algorithm prevents premature convergence by modifying the genetic algorithm and turning it into a multimodal mechanism, while preserving population diversity. The test results of the proposed algorithm on different datasets show that this method improves the accuracy of finding the most influential people in the issue of influence maximization compared to other common algorithms. In social networks, related people are influenced by each other because people share their ideas and views on different issues. If a number of people in a network adopt a particular behavior or belief, this behavior or belief is spread in the network due to the social relationships within the network. This phenomenon is called Influence Propagation. One of the most important issues in influence propagation optimization is the issue of influence maximization. In influence maximization, the goal is to find k subset of members of the social network so that by activating them, under an information diffusion model, the largest number of network members will be affected by information. The purpose of this study is to provide a solution to find the most influential people using the fitness sharing algorithm with dynamic sharing radius and under the Linear Threshold Model. The proposed solution is one of the meta-heuristic solutions in which the genetic algorithm is used. The proposed algorithm prevents premature convergence by modifying the genetic algorithm and turning it into a multimodal mechanism, while preserving population diversity. The test results of the proposed algorithm on different datasets show that this method improves the accuracy of finding the most influential people in the issue of influence maximization compared to other common algorithms. /downloadfilepdf/323218