Scholarlink

<p>In the recent years, the field of recommendation systems has seen significant advancements through the integration of deep reinforcement learning (DRL) and sophisticated text processing techniques. Research work have been proposed a novel framework “Enhanced DQN-Proximal Policy Optimization (PPO) with Text Embedding” which addresses the “cold start” problem—a common issue where the system struggles to make accurate and adequate recommendations about their best preferences, leading to less personalized and potentially irrelevant suggestions especially in case of new users or items due to a lack of insufficient initial data. This proposed work focuses ongoing development of intelligent recommendation systems, provides a robust solution to the cold start problem through the synergy of DRL and text processing. This proposed dual-strategy not only enhances the system’s ability to assists personalized recommendations but also ensures adaptability in dynamic environments. Experiments conducted on benchmark datasets with valid implementation which ensures the highly effectiveness proposed model, revealing notable enhancements in recommendation performance rather than conventional techniques.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An Effective Cold Start Recommendation Model for Smart Decision Using Text Infused Deep Reinforcement Learning

  • 期刊论文

  • 发表日期：

    2026年5月27日

查看全文
  • G Gnanakumar,
  • Vijayakumar Adaickalam

摘要

In the recent years, the field of recommendation systems has seen significant advancements through the integration of deep reinforcement learning (DRL) and sophisticated text processing techniques. Research work have been proposed a novel framework “Enhanced DQN-Proximal Policy Optimization (PPO) with Text Embedding” which addresses the “cold start” problem—a common issue where the system struggles to make accurate and adequate recommendations about their best preferences, leading to less personalized and potentially irrelevant suggestions especially in case of new users or items due to a lack of insufficient initial data. This proposed work focuses ongoing development of intelligent recommendation systems, provides a robust solution to the cold start problem through the synergy of DRL and text processing. This proposed dual-strategy not only enhances the system’s ability to assists personalized recommendations but also ensures adaptability in dynamic environments. Experiments conducted on benchmark datasets with valid implementation which ensures the highly effectiveness proposed model, revealing notable enhancements in recommendation performance rather than conventional techniques.