In recent years, social networks have become very popular and an integral part of everyday life. People express their feelings and experiences in this virtual environment and become aware of others’ opinions and interests. Among them, influential users play an important role in disseminating information on social networks. Identifying such influencers is important in designing techniques to increase the speed of information dissemination. Such techniques are applicable in various fields including viral marketing, preventing the dissemination of harmful information, providing specialized recommendations, etc. Various approaches have been used to detect influencers on social networks, mostly based on the individual’s position in the network structure and their interactions. Considering the strengths and weaknesses of the previous methods, this study presents a novel method based on the content of the users’ posts without considering the network structure. This is done using a combination of high-level features extracted from images to measure the individual’s influence. Users’ images are investigated from three aspects: (1) color scheme, (2) advertising nature, (3) images’ semantics. To describe each of these aspects, feature extraction methods were used with acceptable accuracy in recognizing influential users. Finally, to achieve greater efficiency and precision, feature-combination methods have been investigated to provide an integrated classifier.

Keywords