Jin Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Jin Zhang has a total of 15 co-authors. Here are the top co-authors by number of shared publications:
- Jibing Gong
Yanshan University, Qinhuangdao, China and Hebei Key Laboratory of Computer Virtual Technology and System Integration, Qinhuangdao, China
2 shared publications - Chenglong Wang
School of Information Science and Engineering, Yanshan University, China
1 shared publication - Feng Tian
Northeast Petroleum University
1 shared publication - Gang Huang
Zhejiang University, Hangzhou, China
1 shared publication - Jianjun Xu
Beijing Caizhi Technology Co.,Ltd, China
1 shared publication - Jianqing Li
Macau University of Science and Technology
1 shared publication
Latest Publications
- Open Access
- 14 July 2023
- Published by Association for Computing Machinery (ACM)
- p. 293-300
Author Name Disambiguation based on Capsule Network via Semantic and Structural Features
Author Name Disambiguation (AND) is a crucial task in the knowledge engineering of the bibliography. In academic search systems, author name ambiguity is a common phenomenon caused by different authors with the same name and leads to that author name can not be used to reliably identify all scholar authors. In recent researches, one papers’ attributes are often used to learn its representation as feature. However, most existing methods ignore to extract deep features and potential relationship among papers. To address the problem, we propose a novel model named Author Name Disambiguation based on Capsule Network via Semantic and Structural Features (ADSSF). ADSSF uses both supervised and unsupervised methods to learn the representation of papers. First, we present a new Capsule-Networks-based feature extraction model which can mine deep features and potential relationship. And then, in the representation learning of papers, ADFFS fuses the semantic and structural features of papers by multi-task learning. Finally, a clustering method is leveraged to correctly cluster authors. Experimental results on the AMiner datasets demonstrate that the ADSSF outperforms the state-of-the-art baselines.
- 14 July 2023
- Published by Association for Computing Machinery (ACM)
- p. 293-300
Open AccessAuthor Name Disambiguation based on Capsule Network via Semantic and Structural Features
Author Name Disambiguation (AND) is a crucial task in the knowledge engineering of the bibliography. In academic search systems, author name ambiguity is a common phenomenon caused by different authors with the same name and leads to that author name can not be used to reliably identify all scholar authors. In recent researches, one papers’ attributes are often used to learn its representation as feature. However, most existing methods ignore to extract deep features and potential relationship among papers. To address the problem, we propose a novel model named Author Name Disambiguation based on Capsule Network via Semantic and Structural Features (ADSSF). ADSSF uses both supervised and unsupervised methods to learn the representation of papers. First, we present a new Capsule-Networks-based feature extraction model which can mine deep features and potential relationship. And then, in the representation learning of papers, ADFFS fuses the semantic and structural features of papers by multi-task learning. Finally, a clustering method is leveraged to correctly cluster authors. Experimental results on the AMiner datasets demonstrate that the ADSSF outperforms the state-of-the-art baselines.
- Open Access
- 1 September 2021
- Published by IOP Publishing in Journal of Physics: Conference Series
- Vol. 2026 (1) , 012008
Classification of Channel Coding Structure Based on 1-D Inception Architecture
- 1 September 2021
- Published by IOP Publishing in Journal of Physics: Conference Series
- Vol. 2026 (1) , 012008
Open AccessClassification of Channel Coding Structure Based on 1-D Inception Architecture
- Open Access
- 31 August 2021
- Published by MDPI AG in Algorithms
- Vol. 14 (9) , 263
Sequential Recommendation through Graph Neural Networks and Transformer Encoder with Degree Encoding
Predicting users’ next behavior through learning users’ preferences according to the users’ historical behaviors is known as sequential recommendation. In this task, learning sequence representation by modeling the pairwise relationship between items in the sequence to capture their long-range dependencies is crucial. In this paper, we propose a novel deep neural network named graph convolutional network transformer recommender (GCNTRec). GCNTRec is capable of learning effective item representation in a user’s historical behaviors sequence, which involves extracting the correlation between the target node and multi-layer neighbor nodes on the graphs constructed under the heterogeneous information networks in an end-to-end fashion through a graph convolutional network (GCN) with degree encoding, while the capturing long-range dependencies of items in a sequence through the transformer encoder model. Using this multi-dimensional vector representation, items related to a user historical behavior sequence can be easily predicted. We empirically evaluated GCNTRec on multiple public datasets. The experimental results show that our approach can effectively predict subsequent relevant items and outperforms previous techniques.
- 31 August 2021
- Published by MDPI AG in Algorithms
- Vol. 14 (9) , 263
Open AccessSequential Recommendation through Graph Neural Networks and Transformer Encoder with Degree Encoding
Predicting users’ next behavior through learning users’ preferences according to the users’ historical behaviors is known as sequential recommendation. In this task, learning sequence representation by modeling the pairwise relationship between items in the sequence to capture their long-range dependencies is crucial. In this paper, we propose a novel deep neural network named graph convolutional network transformer recommender (GCNTRec). GCNTRec is capable of learning effective item representation in a user’s historical behaviors sequence, which involves extracting the correlation between the target node and multi-layer neighbor nodes on the graphs constructed under the heterogeneous information networks in an end-to-end fashion through a graph convolutional network (GCN) with degree encoding, while the capturing long-range dependencies of items in a sequence through the transformer encoder model. Using this multi-dimensional vector representation, items related to a user historical behavior sequence can be easily predicted. We empirically evaluated GCNTRec on multiple public datasets. The experimental results show that our approach can effectively predict subsequent relevant items and outperforms previous techniques.
- Open Access
- 1 January 2018
- Published by AIP Publishing in AIP Conference Proceedings
- Vol. 1967 (1) , 020016
Research on distributed heterogeneous data PCA algorithm based on cloud platform
Principal component analysis (PCA) of heterogeneous data sets can solve the problem that centralized data scalability is limited. In order to reduce the generation of intermediate data and error components of distributed heterogeneous data sets, a principal component analysis algorithm based on heterogeneous data sets under cloud platform is proposed. The algorithm performs eigenvalue processing by using Householder tridiagonalization and QR factorization to calculate the error component of the heterogeneous database associated with the public key to obtain the intermediate data set and the lost information. Experiments on distributed DBM heterogeneous datasets show that the model method has the feasibility and reliability in terms of execution time and accuracy.
- 1 January 2018
- Published by AIP Publishing in AIP Conference Proceedings
- Vol. 1967 (1) , 020016
Open AccessResearch on distributed heterogeneous data PCA algorithm based on cloud platform
Principal component analysis (PCA) of heterogeneous data sets can solve the problem that centralized data scalability is limited. In order to reduce the generation of intermediate data and error components of distributed heterogeneous data sets, a principal component analysis algorithm based on heterogeneous data sets under cloud platform is proposed. The algorithm performs eigenvalue processing by using Householder tridiagonalization and QR factorization to calculate the error component of the heterogeneous database associated with the public key to obtain the intermediate data set and the lost information. Experiments on distributed DBM heterogeneous datasets show that the model method has the feasibility and reliability in terms of execution time and accuracy.