Scholarlink

<p>This study introduces a collaborative filtering recommendation algorithm named CFR-FD, designed to tackle common challenges in traditional recommendation systems, particularly the cold start problem and data sparsity issue. By combining fine-grained mining with a neighborhood awareness attention mechanism, the algorithm deeply analyzes user behaviors and project attributes while dynamically focusing on neighborhood information. This dual approach not only enhances the understanding of user preferences and project characteristics but also significantly improves the accuracy and personalization of recommendations. Through experimental validation across three real-world network datasets, the CFR-FD algorithm demonstrates superior performance in recommendation effectiveness and precision compared to existing methods. The proposed solution not only addresses the limitations of traditional algorithms but also offers a more robust and efficient approach for network information recommendations, thereby advancing the capability of recommendation systems to handle sparse data and cold start scenarios effectively.</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

Collaborative filtering recommendation algorithm based on fine-grained mining and neighborhood awareness attention

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月17日

查看全文
  • Jin Jun,
  • Yingyu Li

摘要

This study introduces a collaborative filtering recommendation algorithm named CFR-FD, designed to tackle common challenges in traditional recommendation systems, particularly the cold start problem and data sparsity issue. By combining fine-grained mining with a neighborhood awareness attention mechanism, the algorithm deeply analyzes user behaviors and project attributes while dynamically focusing on neighborhood information. This dual approach not only enhances the understanding of user preferences and project characteristics but also significantly improves the accuracy and personalization of recommendations. Through experimental validation across three real-world network datasets, the CFR-FD algorithm demonstrates superior performance in recommendation effectiveness and precision compared to existing methods. The proposed solution not only addresses the limitations of traditional algorithms but also offers a more robust and efficient approach for network information recommendations, thereby advancing the capability of recommendation systems to handle sparse data and cold start scenarios effectively.