The role of opinion in influence maximization in online social networks

Abstract

Influence Maximization (IM) is an optimization problem in social networking to identify the most influential users in social networks. IM problem, with large-scale data, faces many challenges like time-efficiency, accuracy, and effectiveness of seed. To solve these challenges, we propose a Community-based Context-aware Influence Maximization algorithm. Community-based Context-aware Influence Maximization uses a community-based framework to improve time efficiency, significantly reducing search space. It considers the user’s interests to address the effectiveness of the seed. The traditional information diffusion models (i.e., lndependent cascade model) have been extended to the Context-aware Context-aware Independent Cascade model for influence spreading. To identify the most influential users, first propose a Community Detection Algorithm (CDA) to partition the network into sub-networks based on the community label. After analyzing the dataset, a total of 3 communities BJP, TMC, and Sanjukta Morcha are formed. First, the Greedy approach implements the IM algorithm and then the CELF approach. Then the comparison result of the execution time of each algorithm is represented by the Python module matplotlib. After finding the result on the main dataset the Context-aware Graph Reduce Algorithm (Section: 4.4) is used to reduce the size of the main graph. After graph reduction, the IM algorithm (Section: 4.1) with the Greedy approach (Section: 4.2) and the CELF approach (Section: 4.3), and then the comparison result of both the algorithms is represented by the Python package matplotlib. So, this paper highlights how the context-sensitive framework enables faster influence maximization over the network using existing Greedy and CELF models of influence spread.