To calculate the propagation probability of seed users and provide accurate advertising placement locations for merchants, research on location promotion algorithms has attracted widespread attention. Location-based social networks (LBSN) have the ability to quickly extract users’ information. Although they can accurately screen seed users, they do not take into account the personal factors of users, resulting in low prediction accuracy of propagation probability. This study is based on candidate seed selection (CSS) algorithms for check-in prediction using the help dataset, introducing user check-in similarity and spreader influence to optimize CSS algorithms and generate fusion algorithms. Research is conducted to establish a merchant-feature-integrated cascade model (MFICM) that combines attributes such as user check-in information and merchant prosperity to improve the prediction accuracy of user propagation. The dataset with check-in similarity is incorporated to avoid selecting isolated users as active seeds and the influence of disseminators is considered to address the impact of user personal behaviour on the results, focusing on active seeds, which are users who are successfully spreading information or behaviours to others within their social network. Finally, the algorithm studied is applied to the help dataset and the performance of the particle swarm optimization algorithm (PSO) is tested and compared with the research. A total of 500 experiments were conducted in the study. Among the 392 accurate prediction experiments of the fusion algorithm, personal factors accounted for 81.9% and merchant factors accounted for 18.1%. The experimental results indicated that the behavioural attributes of seed users had a direct impact on their check-in behaviour and the promotion position of advertisements should be based on the seed users. • USCI-CSS facilitates seed selection in the integration of user behavior and social influence. • USS, CI and CSS improve the prediction and target advertising accuracy for check-in, target marketing, even better. • The USCI-CSS produces more accurate seed propagation accuracy compared to PSO in prediction rates. • The USCI-CSS adapts to community sizes so that seed propagation can occur better. • Merchant factors such as cleanliness and price can help select seeds for advertising. • MFICM helps us identify check-ins using user behavior and environmental conditions.
Shiming Jiang (Sun,) studied this question.