Bin Yang
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
Research Subjects
Bin Yang has a total of 29 co-authors. Here are the top co-authors by number of shared publications:
Wei Li
13 shared publications- Bao Qing Hu
School of Mathematical Sciences, Center for Applied Mathematics of Guangxi, Guangxi Minzu University, Nanning, People’s Republic of China
5 shared publications Yuanhao Liu
College of Science, Northwest A &F University, YangLing, PR China
4 shared publicationsChengwei Liao
3 shared publicationsHaohao Wang
College of Science, Northwest A &F University, Yangling, People’s Republic of China
3 shared publicationsShuai Liu
3 shared publications
Latest Publications
- Open Access
- 16 September 2026
- Published by Elsevier in SSRN Electronic Journal
Multi-label feature selection with optimistic and pessimistic advantageous labels
Reducing the feature dimension while keeping classification accuracy acceptable is a central concern in multi-label learning. Running single-label selection methods repeatedly does not solve the problem, since label dependence and feature redundancy make the multi-label case far more complicated. Rough set theory and fuzzy set theory are well suited to data described by multiple labels, yet turning the uncertainty they model into an explicit characterization of what fuzzy rough sets can discriminate is still unsettled. Motivated by this gap, we start from distinguishable sets and define the optimistic advantageous labels and pessimistic advantageous labels. These two label sets are then combined with the parameterized fuzzy decision and the inclusion relation to build the optimistic and pessimistic multi-label variable precision fuzzy rough set models, from which two multi-label feature selection algorithms are derived. Experiments on eight datasets support the effectiveness of the proposed methods.
- 16 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessMulti-label feature selection with optimistic and pessimistic advantageous labels
Reducing the feature dimension while keeping classification accuracy acceptable is a central concern in multi-label learning. Running single-label selection methods repeatedly does not solve the problem, since label dependence and feature redundancy make the multi-label case far more complicated. Rough set theory and fuzzy set theory are well suited to data described by multiple labels, yet turning the uncertainty they model into an explicit characterization of what fuzzy rough sets can discriminate is still unsettled. Motivated by this gap, we start from distinguishable sets and define the optimistic advantageous labels and pessimistic advantageous labels. These two label sets are then combined with the parameterized fuzzy decision and the inclusion relation to build the optimistic and pessimistic multi-label variable precision fuzzy rough set models, from which two multi-label feature selection algorithms are derived. Experiments on eight datasets support the effectiveness of the proposed methods.
- 18 July 2026
- Published by Elsevier in Pattern Recognition
- Vol. 180
Kernelized fuzzy β-covering rough set and its application to noise-resilient multilabel feature selection
- 18 July 2026
- Published by Elsevier in Pattern Recognition
- Vol. 180
Kernelized fuzzy β-covering rough set and its application to noise-resilient multilabel feature selection
- Open Access
- 4 June 2026
- Published by Elsevier in SSRN Electronic Journal
A personalized recommendation model based on online reviews: an attribute-weighted random walk with restart approach
Online shopping platforms play a role in consumers' purchase decision-making processes. With the advancement of information technology, product evaluation has evolved beyond numerical ratings, as consumers rely on online reviews containing detailed attribute-level information. However, extant studies often overlook the heterogeneity in consumers' attribute preferences, thereby failing to satisfy consumers' actual needs and to capture their psychological behaviors in online shopping contexts. To address these limitations, this paper proposes a personalized recommendation model based on online reviews, which generates differentiated competitive product ranking lists for distinct consumer groups. Specifically, the attribute-enhanced sentiment analysis bidirectional encoder representations from transformers (AttrSA-BERT) model is first employed to conduct aspect-based sentiment analysis (ABSA) on review texts to obtain multi-attribute sentiment scores. Subsequently, the attribute aware personalized ranking model (AAPRM) framework is utilized to produce competitive rankings for target products, where in a weighted Euclidean distance function is applied for users with predominantly positive reviews, while an improved random walk with restart (IRWR) algorithm is adopted for users with predominantly negative reviews. Finally, a case study based on 57,129 online reviews collected from JD.com is conducted to validate the effectiveness of the proposed approach. The experimental results demonstrate that AttrSA-BERT outperforms all baseline models in terms of evaluation metrics such as accuracy. In addition, the case studies further validate the rationality and effectiveness of the proposed personalized recommendation model, AAPRM. Compared with classical multi-criteria decision-making methods, the proposed model achieves the best performance in terms of both Spearman's rank correlation coefficient and Kendall's tau coefficient.
- 4 June 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessA personalized recommendation model based on online reviews: an attribute-weighted random walk with restart approach
Online shopping platforms play a role in consumers' purchase decision-making processes. With the advancement of information technology, product evaluation has evolved beyond numerical ratings, as consumers rely on online reviews containing detailed attribute-level information. However, extant studies often overlook the heterogeneity in consumers' attribute preferences, thereby failing to satisfy consumers' actual needs and to capture their psychological behaviors in online shopping contexts. To address these limitations, this paper proposes a personalized recommendation model based on online reviews, which generates differentiated competitive product ranking lists for distinct consumer groups. Specifically, the attribute-enhanced sentiment analysis bidirectional encoder representations from transformers (AttrSA-BERT) model is first employed to conduct aspect-based sentiment analysis (ABSA) on review texts to obtain multi-attribute sentiment scores. Subsequently, the attribute aware personalized ranking model (AAPRM) framework is utilized to produce competitive rankings for target products, where in a weighted Euclidean distance function is applied for users with predominantly positive reviews, while an improved random walk with restart (IRWR) algorithm is adopted for users with predominantly negative reviews. Finally, a case study based on 57,129 online reviews collected from JD.com is conducted to validate the effectiveness of the proposed approach. The experimental results demonstrate that AttrSA-BERT outperforms all baseline models in terms of evaluation metrics such as accuracy. In addition, the case studies further validate the rationality and effectiveness of the proposed personalized recommendation model, AAPRM. Compared with classical multi-criteria decision-making methods, the proposed model achieves the best performance in terms of both Spearman's rank correlation coefficient and Kendall's tau coefficient.
- Open Access
- 23 February 2026
- Published by Elsevier in SSRN Electronic Journal
Evaluating green low-carbon transformation performance of climate-resilient cities via a multi-criteria evaluation method with multiple reference points
Evaluating the green low-carbon transformation performance of climate-resilient pilot cities is essential for promoting sustainable development under global climate change in China. Nevertheless, existing multi-criteria evaluation methods often lack temporal consistency in criterion weight determination and setting reference points, limiting their applicability to multi-stage assessments. To address these issues, this paper develops an improved time-integrated weighting method based on integrated determination of objective criteria weights and intuitionistic fuzzy entropy, which considers both information utility and relative information loss over the time period. Furthermore, grey relational analysis is introduced to capture intertemporal correlations between criteria. Accordingly, we establish a multi-stage multi-criteria evaluation model that integrates global and stage-specific reference points with regret utility adjustments. The proposed model is applied to 21 climate-resilient pilot cities in China from 2019 to 2023, with sensitivity analysis conducted to verify its validity. Empirical findings indicate that Chongqing ranks first in overall transformation performance throughout the study period. Main driving factors include residents' income, population size, and green-space coverage, while low-carbon criteria have limited contributions.
- 23 February 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessEvaluating green low-carbon transformation performance of climate-resilient cities via a multi-criteria evaluation method with multiple reference points
Evaluating the green low-carbon transformation performance of climate-resilient pilot cities is essential for promoting sustainable development under global climate change in China. Nevertheless, existing multi-criteria evaluation methods often lack temporal consistency in criterion weight determination and setting reference points, limiting their applicability to multi-stage assessments. To address these issues, this paper develops an improved time-integrated weighting method based on integrated determination of objective criteria weights and intuitionistic fuzzy entropy, which considers both information utility and relative information loss over the time period. Furthermore, grey relational analysis is introduced to capture intertemporal correlations between criteria. Accordingly, we establish a multi-stage multi-criteria evaluation model that integrates global and stage-specific reference points with regret utility adjustments. The proposed model is applied to 21 climate-resilient pilot cities in China from 2019 to 2023, with sensitivity analysis conducted to verify its validity. Empirical findings indicate that Chongqing ranks first in overall transformation performance throughout the study period. Main driving factors include residents' income, population size, and green-space coverage, while low-carbon criteria have limited contributions.
- 1 January 2026
- Published by Tech Science Press in Computers, Materials & Continua
- p. 1-10
An Online Review-Driven Large-Scale Group Decision-Making Approach for User Preference Evaluation Considering Individual Delayed Response Behavior
Online reviews provide valuable insights into user preferences. However, the uncertainty inherent in expressed opinions and delayed responses by decision-makers can complicate consensus formation in large groups. To address these challenges, this study proposes an online review-driven, large-scale group decision-making framework to prioritize key factors associated with user satisfaction in fresh fruit e-commerce. The core of this study is a cumulative prospect theory (CPT)-based model of decision-maker delayed response behavior in a picture fuzzy environment. Picture fuzzy sets represent uncertainty in review-derived opinions, while the Lancichinetti-Fortunato method identifies overlapping subgroups and preserves individuals’ cross-subgroup bridging roles in social networks. Using subgroup preferences as reference points, the model links perceived gains and losses to decision-makers’ acceptance or postponement of preference adjustments in the consensus reaching process. A multi-round feedback mechanism integrates CPT-based response prediction with genetic algorithm (GA) optimization to identify cost-efficient preference adjustments that satisfy the prescribed consensus threshold. A case study based on 29 fresh fruit stores on an e-commerce platform identifies logistics and product quality as the two highest-ranked satisfaction factors. Comparative analyses with benchmark methods show that the proposed model achieves lower adjustment costs while meeting the same consensus threshold and yielding an identical ranking, supporting its ranking robustness.
- 1 January 2026
- Published by Tech Science Press in Computers, Materials & Continua
- p. 1-10
An Online Review-Driven Large-Scale Group Decision-Making Approach for User Preference Evaluation Considering Individual Delayed Response Behavior
Online reviews provide valuable insights into user preferences. However, the uncertainty inherent in expressed opinions and delayed responses by decision-makers can complicate consensus formation in large groups. To address these challenges, this study proposes an online review-driven, large-scale group decision-making framework to prioritize key factors associated with user satisfaction in fresh fruit e-commerce. The core of this study is a cumulative prospect theory (CPT)-based model of decision-maker delayed response behavior in a picture fuzzy environment. Picture fuzzy sets represent uncertainty in review-derived opinions, while the Lancichinetti-Fortunato method identifies overlapping subgroups and preserves individuals’ cross-subgroup bridging roles in social networks. Using subgroup preferences as reference points, the model links perceived gains and losses to decision-makers’ acceptance or postponement of preference adjustments in the consensus reaching process. A multi-round feedback mechanism integrates CPT-based response prediction with genetic algorithm (GA) optimization to identify cost-efficient preference adjustments that satisfy the prescribed consensus threshold. A case study based on 29 fresh fruit stores on an e-commerce platform identifies logistics and product quality as the two highest-ranked satisfaction factors. Comparative analyses with benchmark methods show that the proposed model achieves lower adjustment costs while meeting the same consensus threshold and yielding an identical ranking, supporting its ranking robustness.
- 9 August 2025
- Published by Elsevier in Applied Soft Computing
- Vol. 184
A two-stage feature selection approach with fuzzy covering-based rough sets based on discernibility matrix
- 9 August 2025
- Published by Elsevier in Applied Soft Computing
- Vol. 184
A two-stage feature selection approach with fuzzy covering-based rough sets based on discernibility matrix
- 16 April 2025
- Published by Elsevier in Fuzzy Sets and Systems
- Vol. 515
Optimal scale selection of dynamic incomplete generalized multi-scale fuzzy ordered decision systems based on rough fuzzy sets
- 16 April 2025
- Published by Elsevier in Fuzzy Sets and Systems
- Vol. 515
Optimal scale selection of dynamic incomplete generalized multi-scale fuzzy ordered decision systems based on rough fuzzy sets
- 25 March 2025
- Published by Elsevier in Fuzzy Sets and Systems
- Vol. 512
A novel method for incremental feature selection with fuzzy β-covering
- 25 March 2025
- Published by Elsevier in Fuzzy Sets and Systems
- Vol. 512
A novel method for incremental feature selection with fuzzy β-covering