Article
Computer Science, Artificial Intelligence
Yuntao Du, Xinjun Zhu, Lu Chen, Ziquan Fang, Yunjun Gao
Summary: A knowledge graph (KG) is a set of interconnected entities and attributes, which can be used as auxiliary information for accurate, explainable, and diverse user preference recommendations. However, existing KG recommendation methods often ignore cold-start problems, resulting in poor performance when dealing with new users or items. In this paper, we propose a meta-learning based framework called MetaKG, which includes collaborative-aware and knowledge-aware meta learners to address cold-start recommendations. Experimental results demonstrate that MetaKG outperforms existing state-of-the-art methods in terms of effectiveness, efficiency, and scalability.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Computer Science, Artificial Intelligence
Renchu Guan, Haoyu Pang, Fausto Giunchiglia, Yanchun Liang, Xiaoyue Feng
Summary: The cold-start problem limits the effectiveness of recommendation systems. There are two main strategies to address this problem: cross-domain recommendation (CDR) and meta-learning. However, CDR methods lack optimization for the few-shot problem, while most meta-learning approaches ignore cross-domain information. Therefore, a novel approach called MetaCDR is proposed, which combines domain knowledge and meta-optimization to solve the cold-start problem.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Computer Science, Artificial Intelligence
Jia Xu, Hongming Zhang, Xin Wang, Pin Lv
Summary: To address the problem of user cold-start recommendation, a novel adaptive meta-learning model based on user relevance (AdaML) is proposed. This model identifies related users with similar preferences and utilizes their information to improve user cold-start recommendations.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Huiting Liu, Lei Wang, Peipei Li, Cheng Qian, Peng Zhao, Xindong Wu
Summary: To address the cold-start problem and improve the generalization performance of meta-learning, we propose a relation-propagation meta-learning method on explicit preference graph. Our method captures the relationships between local preferences and produces more distinguishable local preference nodes using graph convolutional networks. Experimental results demonstrate its effectiveness.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Computer Science, Information Systems
Honglin Shu, Fu -Lai Chung, Da Lin
Summary: The MetaGC-MC approach, based on graph convolutional networks and meta-learning, provides effective cold-start recommendations without relying on auxiliary data. By random sampling subgraphs as meta-learning tasks, it captures various subgraph structure information and encodes it as a meta-prior for rapid adaptions. MetaGC-MC can also utilize auxiliary data to enhance model performance.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Information Systems
Joojo Walker, Fengli Zhang, Ting Zhong, Fan Zhou, Edward Yellakuor Baagyere
Summary: In this paper, a recommendation model called CORE-VAE is proposed to address the complexity and sparsity of social networks by utilizing a social-aware similarity function and a graph convolutional network. The model generates cold-start resistant rating vectors by producing robust social-aware user representations and explores user rating information using an expressive variational autoencoder. Experimental results demonstrate that CORE-VAE outperforms competitive models on real-world datasets.
INFORMATION SCIENCES
(2022)
Article
Computer Science, Artificial Intelligence
Israr ur Rehman, Waqar Ali, Zahoor Jan, Zulfiqar Ali, Hui Xu, Jie Shao
Summary: The performance of recommendation engines is challenged by cold-start issues. Traditional recommendation techniques have limited capability to address this problem due to data-thirsty machine learning models. On the contrary, the recent success of meta-learning, with its few-shot learning capabilities, has attracted attention for accommodating new tasks with few data samples. We propose a Contextually Augmented Meta-Learning recommender system (CAML) to improve task adaptation capability by augmenting contextual features into a meta-learning model.
Article
Computer Science, Information Systems
Yunyi Li, Yongjing Hao, Pengpeng Zhao, Guanfeng Liu, Yanchi Liu, Victor S. Sheng, Xiaofang Zhou
Summary: This article proposes an Edge-Enhanced Global Disentangled Graph Neural Network (EGD-GNN) model to capture the relation information between items for global item representation and local user intention learning. Experimental results show that our model can achieve a significant improvement over state-of-the-art baselines and effectively distinguish item features.
ACM TRANSACTIONS ON KNOWLEDGE DISCOVERY FROM DATA
(2023)
Article
Computer Science, Artificial Intelligence
Tianyuan Li, Xin Su, Wei Liu, Wei Liang, Meng-Yen Hsieh, Zhuhui Chen, XuChong Liu, Hong Zhang
Summary: Personalised recommendation is a challenging issue, especially cold-start recommendation due to sparse user-item interaction. This paper proposed a memory-augmented meta-learning method on the meta-path to address cold-start recommendation, showing efficiency in experiments with widely used datasets and cold-start scenarios.
CONNECTION SCIENCE
(2022)
Article
Computer Science, Artificial Intelligence
Xiaolin Zheng, Yanchao Tan, Yan Wang, Xiangyu Wei, Shengjia Zhang, Chaochao Chen, Longfei Li, Carl Yang
Summary: The sparse interactions between users and items on the web have made it difficult to represent them in recommender systems. Existing approaches use item attributes to alleviate the data sparsity problem, but manual labeling of attribute quality is time-consuming. In response, HQRec is proposed to automatically measure attribute quality and make accurate recommendations. HQRec achieves significant performance gains over state-of-the-art baselines, with an average improvement of 14.73% in terms of Recall and NDCG metrics.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Computer Science, Artificial Intelligence
Yan Zhang, Ivor W. Tsang, Hongzhi Yin, Guowu Yang, Defu Lian, Jingjing Li
Summary: This paper proposes a Deep Pairwise Hashing (DPH) method to map users and items to binary vectors in Hamming space, which efficiently calculates a user's preference for an item. To address data sparsity and cold-start problems, user-item interactive information and item content information are combined to learn effective representations. DPH achieves significant improvement in data sparsity and item cold-start recommendation compared to state-of-the-art frameworks.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2022)
Article
Automation & Control Systems
Jinjin Zhang, Chenhui Ma, Chengliang Zhong, Peng Zhao, Xiaodong Mu
Summary: This paper proposes a novel framework called FINI, which utilizes feature weights and interactions between neighbor nodes to improve cold start recommendation. By designing a global-local contexts attention mechanism and a mixed interaction mechanism, the expressive capability of feature embeddings and user/item embeddings are enhanced, leading to significant improvements in terms of metric evaluations.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2022)
Article
Computer Science, Artificial Intelligence
Tieyun Qian, Yile Liang, Qing Li, Hui Xiong
Summary: Rating prediction is a classic problem addressed by matrix factorization, but recent advancements in deep learning, particularly graph neural networks, have shown impressive progress. This study introduces a new AGNN framework that utilizes attribute graphs to learn preference embeddings for strict cold start users/items.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2022)
Article
Computer Science, Information Systems
Chunyang Wang, Yanmin Zhu, Haobing Liu, Tianzi Zang, Ke Wang, Jiadi Yu
Summary: In this article, we propose a multifaceted relation-aware meta-learning approach called MeCM for user cold-start recommendation. This approach enhances task-adaptive initialization customization by extracting multiple views of task relevance. Extensive experiments demonstrate that MeCM outperforms state-of-the-art meta-learning-based recommendation methods.
ACM TRANSACTIONS ON KNOWLEDGE DISCOVERY FROM DATA
(2023)
Article
Computer Science, Information Systems
Desheng Cai, Shengsheng Qian, Quan Fang, Jun Hu, Changsheng Xu
Summary: Recently, there has been a focus on user cold-start recommendations in both industry and academia. Existing approaches often overlook the sparsity of user attributes in these systems. To address this limitation, this article proposes a novel Inductive Heterogeneous Graph Neural Network (IHGNN) model that utilizes relational information to alleviate attribute sparsity. Experimental results demonstrate that the IHGNN outperforms existing baselines.
ACM TRANSACTIONS ON INFORMATION SYSTEMS
(2023)