فهرست نشریات علمی دسترسی آزاد ایران

Social networks such as Twitter serve as influential platforms for executing campaigns aimed at increasing user engagement. These campaigns typically involve large-scale, topic-focused interactions among users over consecutive time periods. Achieving successful outcomes depends on two challenges: accurately identifying appropriate target audiences and delivering tailored content. To address these issues, we propose a novel framework that combines recommender systems with large language models (LLMs). While existing approaches often rely on manual campaign execution without systematic targeting, our method integrates collaborative and content-based filtering with LLMs to recommend relevant content and identify suitable users for bot-controlled accounts. The framework analyzes users’ behavioral feedback to infer preferences and evaluates tweet content for semantic relevance, enabling the system to deliver impactful messages and effectively identify target audiences. Campaigns are initiated within a controlled set of bot-controlled user accounts, which simulate human behaviors (e.g., posting, retweeting), establishing baseline engagement patterns for systematic evaluation of the recommender system’s performance before expanding to human audiences. Our evaluation focuses on two key metrics: hashtag adoption rate and user engagement. Comparative experiments demonstrate that our hybrid approach significantly enhances both metrics, confirming its effectiveness in improving campaign outcomes through targeted content delivery and automated user engagement strategies.