AI-Driven Market Segmentation and Multi-Behavioral Sequential Recommendation for Personalized E-Commerce Marketing

In the era of AI-driven e-commerce and advertising platforms, market segmentation and personalized recommendation have become essential for improving user conversion rates and marketing effectiveness. By leveraging artificial intelligence to conduct deep analysis of large-scale behavioral data from e-commerce platforms, it is possible to perform precise customer segmentation, identify diverse consumer groups, and develop customized marketing strategies. However, users in real-world recommendation scenarios typically exhibit multiple interaction behaviors—such as clicking, adding to cart, and purchasing—which makes it difficult for traditional single-task models to learn generalized representations without introducing task-specific biases. To address this challenge, we propose a pre-training paradigm designed to decouple task-specific and general knowledge in multi-behavior sequential recommendation (MBSR). Yet, conventional pre-trained models are often too large for practical adaptation by end users. Inspired by the success of prompt learning in the natural language processing field, we introduce CPL4Rec (Customized Prompt Learning for Recommendation), the first framework for customized prompt learning in MBSR. CPL4Rec generates user-specific prompts by integrating semantic embeddings from pre-trained models with diverse user attributes and behavioral information. Furthermore, to address the evolving nature of user interests over time, we incorporate a Progressive Feature Generation (PFG) framework that dynamically fuses multi-layer user representations within the model. To ensure controllability, we apply compactness regularization to constrain the prompt space. Extensive experiments conducted on three real-world datasets demonstrate that CPL4Rec achieves superior performance over state-of-the-art baselines in recommendation accuracy. This research offers a new technical pathway for AI-driven market segmentation and personalized e-commerce marketing, providing strong theoretical and empirical support for practical deployment in intelligent recommendation systems.