Improved information diffusion models based on a new two-sided sign-aware matching framework in complex networks

The information diffusion model is an important trending topic in the influence maximization problem that companies use it aiming to promote effective viral advertising campaigns at a low cost. The majority of existing models still do not fulfill the need for sufficient prediction accuracy in real-world applications. The main reason is that most of the current information diffusion models do not consider some important aspects of complicated relationships among network users. As one of the important users' relationships aspects, there are extensively reciprocal relationships along with parasocial relationships simultaneously based on different trust levels, including trust, mediocre trust, and distrust relationships in almost all real-world complex networks, but these models have not considered them. Motivated by these limitations, in this study, we introduce a two-sided signaware matching (TSM) framework to comprehensively describe the influence propagation process according to two new scenarios in signed social networks (SSNs). Under the TSM framework, we propose four effective information diffusion models: SCSI-B, SCSI-N, STSI-B, and STSI-N, which are divided into the cascade and threshold-based group of models. The experimental results on real-world SSNs showed that the proposed models outperform the state-of-the-art baselines with improvements ranging from 7.4% to 20.8% in terms of the F-score metric according to their corresponding scenarios. In addition, the improvements of proposed models against classic baselines are from 24.9 % to 45 %, and hence they can be used to facilitate intelligent decision-making in viral marketing.