Scholarlink

<p>In the digital age, recommendation systems are essential for guiding users to personalized content. While content-based and collaborative filtering methods are widely used, they often struggle to capture detailed user preferences and item characteristics. This paper explores advanced recommendation techniques, with a focus on two innovative applications, multimodal recommendation systems and knowledge graph-based recommendation systems. By proposing a recommender system that integrates multiple data modalities and leveraging the structured relationships within knowledge graphs, this work aims to provide contextually rich and personalized recommendations. Experimental results show the superiority of the proposed model compared to the benchmark approaches achieving recall values of 0.521, 0.729, and 0.739 across different experiments. These results represent substantial improvements over benchmark models in terms of recall, with increases ranging from 18.9% to 46.8%. The Normalized Discounted Cumulative Gain (NDCG) also saw significant gains, with the proposed model reaching up to 0.880, showcasing enhancements of up to 39.6% over the benchmark models. Rigorous evaluation and assessment validate the proposed model’s ability to deliver robust, contextually rich, and personalized recommendations, demonstrating its effectiveness in addressing common limitations such as the limited user-item interactions. These promising outcomes indicate a future where recommendation systems not only offer greater but also gain a deeper understanding of users’ detailed and specific needs.</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

Multimodal fusion in graph-based recommendation systems

  • 期刊论文

  • 发表日期：

    2026年2月3日

查看全文
  • Maha Sayed,
  • Wedad Hussien,
  • Yasmine M. Afify,
  • Walaa Gad

摘要

In the digital age, recommendation systems are essential for guiding users to personalized content. While content-based and collaborative filtering methods are widely used, they often struggle to capture detailed user preferences and item characteristics. This paper explores advanced recommendation techniques, with a focus on two innovative applications, multimodal recommendation systems and knowledge graph-based recommendation systems. By proposing a recommender system that integrates multiple data modalities and leveraging the structured relationships within knowledge graphs, this work aims to provide contextually rich and personalized recommendations. Experimental results show the superiority of the proposed model compared to the benchmark approaches achieving recall values of 0.521, 0.729, and 0.739 across different experiments. These results represent substantial improvements over benchmark models in terms of recall, with increases ranging from 18.9% to 46.8%. The Normalized Discounted Cumulative Gain (NDCG) also saw significant gains, with the proposed model reaching up to 0.880, showcasing enhancements of up to 39.6% over the benchmark models. Rigorous evaluation and assessment validate the proposed model’s ability to deliver robust, contextually rich, and personalized recommendations, demonstrating its effectiveness in addressing common limitations such as the limited user-item interactions. These promising outcomes indicate a future where recommendation systems not only offer greater but also gain a deeper understanding of users’ detailed and specific needs.