Le Song
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
Le Song has a total of 32 co-authors. Here are the top co-authors by number of shared publications:
- Yinghong Ma
Business School, Shandong Normal University, Jinan, 250014, China
7 shared publications - Xiao Yin
School of Business Administration, South China University of Technology, Guangzhou 510641, China
2 shared publications - Changyu Liu
Department of Statistics, The Chinese University of Hong Kong , Shatin, New Territories, Hong Kong
1 shared publication - Gege Zhan
Laboratory for Neural Interface and Brain Computer Interface, Engineering Research Center of AI & Robotics, Ministry of Education, Shanghai Engineering Research Center of AI & Robotics, MOE Frontiers Center for Brain Science, State Key Laboratory of Medical Neurobiology, Institute of AI & Robotics, Institute of Meta-Medical, Academy for Engineering & Technology, Fudan University, Shanghai, People’s Republic of China
1 shared publication - Haifeng Jiang
Nanjing University of Science and Technology School of Automation
1 shared publication Han Zhou
1 shared publication
Latest Publications
- Open Access
BiLSTM-ResNet-CRF: An Improved Model for Subject Knowledge Graph Construction
The emergence of massive knowledge in online learning systems has increased the difficulty for learners to acquire the necessary information. Due to unclear information expression and excessive knowledge redundancy, learners face challenges in identifying relevant knowledge. Furthermore, the presence of substantial unstructured knowledge in subject domains also hinders the effective transmission and application of knowledge. To address these issues, a framework for constructing a subject domain knowledge graph is proposed in this work. The framework primarily aims to visualize isolated information and connect knowledge into graph structures. The knowledge graph can help learners quickly and efficiently acquire the knowledge they need. The novel framework is constructed with three steps. The first step is to design the ontology rules based on the domain-specific subject knowledge from the perspective of classification, and also to construct the schema layer of the knowledge graph. The second step is to propose a domain-optimized BiLSTM-ResNet-CRF model for subject domain entity recognition, which introduces residual blocks to enhance fine-grained local contextual feature extraction for multi-word technical terms, addressing the limitations of traditional BiLSTM-CRF models in educational text processing. The BERT relation extraction model is used to extract relations between knowledge entities. Then the data layer is constructed. Finally, the third step is to achieve knowledge fusion through entity linking and two-layer entity alignment against results stored in a database. The result comparisons on the dataset show that the novel BiLSTM-ResNet-CRF model has higher scores than several other classical models, achieving an F1-score of 80.26%. The proposed framework’s effectiveness is rigorously validated using high school mathematics as a representative case study with a well-structured knowledge system.
Open AccessBiLSTM-ResNet-CRF: An Improved Model for Subject Knowledge Graph Construction
The emergence of massive knowledge in online learning systems has increased the difficulty for learners to acquire the necessary information. Due to unclear information expression and excessive knowledge redundancy, learners face challenges in identifying relevant knowledge. Furthermore, the presence of substantial unstructured knowledge in subject domains also hinders the effective transmission and application of knowledge. To address these issues, a framework for constructing a subject domain knowledge graph is proposed in this work. The framework primarily aims to visualize isolated information and connect knowledge into graph structures. The knowledge graph can help learners quickly and efficiently acquire the knowledge they need. The novel framework is constructed with three steps. The first step is to design the ontology rules based on the domain-specific subject knowledge from the perspective of classification, and also to construct the schema layer of the knowledge graph. The second step is to propose a domain-optimized BiLSTM-ResNet-CRF model for subject domain entity recognition, which introduces residual blocks to enhance fine-grained local contextual feature extraction for multi-word technical terms, addressing the limitations of traditional BiLSTM-CRF models in educational text processing. The BERT relation extraction model is used to extract relations between knowledge entities. Then the data layer is constructed. Finally, the third step is to achieve knowledge fusion through entity linking and two-layer entity alignment against results stored in a database. The result comparisons on the dataset show that the novel BiLSTM-ResNet-CRF model has higher scores than several other classical models, achieving an F1-score of 80.26%. The proposed framework’s effectiveness is rigorously validated using high school mathematics as a representative case study with a well-structured knowledge system.
- 5 December 2025
- Published by Elsevier in Physica A: Statistical Mechanics and its Applications
- Vol. 682
A node ranking method: Reducing structural indexes to avoid evaluation redundancy
- 5 December 2025
- Published by Elsevier in Physica A: Statistical Mechanics and its Applications
- Vol. 682
A node ranking method: Reducing structural indexes to avoid evaluation redundancy
- Open Access
Knowledge Recombination Reveals the Nonlinear Influence of Team Scale on Technological Breakthroughs
In the knowledge economy era, optimizing R&D team size is crucial for breakthrough innovation. Breakthrough technologies rely more on knowledge restructuring and technological leaps than general technologies do. However, it remains unclear whether breakthrough technology formation follows a simple “more people, more power” logic within technological systems. This work examines 35,955 patents in recommendation system technology to propose a relationship model between collaboration scale and breakthrough technological innovation based on patent data from the recommendation system field. It aims to elucidate how collaboration scale influences breakthrough technological innovation through knowledge restructuring, thereby providing theoretical support and practical guidance for enterprises, institutions, and governments in innovation activities to advance technological innovation. The findings reveal three key points: (1) The relationship between collaboration scale and breakthrough innovation is not linear but follows an inverted U-shaped curve; (2) Knowledge recombination significantly mediates this relationship, also exhibiting an inverted U-shaped pattern with collaboration scale; (3) The inverted U-shaped effect of collaboration scale on breakthrough innovation varies by country. The optimal thresholds are 14.058 entities for China, 57.151 entities for the United States, and 4.801 entities for Russia. This work breaks through the limitations of the traditional theoretical framework and constructs a three-dimensional analysis framework of “collaboration scale → knowledge recombination → breakthrough technological innovation”. By introducing the mediating variable of knowledge recombination, this paper reveals the mechanism of R&D team size on radical innovation. It provides a theoretical basis for the construction of an innovation team and provides a theoretical basis for enterprises, governments, and institutions.
Open AccessKnowledge Recombination Reveals the Nonlinear Influence of Team Scale on Technological Breakthroughs
In the knowledge economy era, optimizing R&D team size is crucial for breakthrough innovation. Breakthrough technologies rely more on knowledge restructuring and technological leaps than general technologies do. However, it remains unclear whether breakthrough technology formation follows a simple “more people, more power” logic within technological systems. This work examines 35,955 patents in recommendation system technology to propose a relationship model between collaboration scale and breakthrough technological innovation based on patent data from the recommendation system field. It aims to elucidate how collaboration scale influences breakthrough technological innovation through knowledge restructuring, thereby providing theoretical support and practical guidance for enterprises, institutions, and governments in innovation activities to advance technological innovation. The findings reveal three key points: (1) The relationship between collaboration scale and breakthrough innovation is not linear but follows an inverted U-shaped curve; (2) Knowledge recombination significantly mediates this relationship, also exhibiting an inverted U-shaped pattern with collaboration scale; (3) The inverted U-shaped effect of collaboration scale on breakthrough innovation varies by country. The optimal thresholds are 14.058 entities for China, 57.151 entities for the United States, and 4.801 entities for Russia. This work breaks through the limitations of the traditional theoretical framework and constructs a three-dimensional analysis framework of “collaboration scale → knowledge recombination → breakthrough technological innovation”. By introducing the mediating variable of knowledge recombination, this paper reveals the mechanism of R&D team size on radical innovation. It provides a theoretical basis for the construction of an innovation team and provides a theoretical basis for enterprises, governments, and institutions.
- Open Access
- 26 June 2025
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 20 (6) , e0326736
Diversity in knowledge search paths and breakthrough technological innovation: Evidence from new energy vehicle firms
Breakthrough technological innovations have heralded a transformative era within the automotive industry. Knowledge search diversity, as the key to acquiring the raw materials for knowledge recombination, enables firms to overcome resource constraints and accelerate the realization of breakthrough technological innovations. This paper integrates the theories of organizational search and knowledge recombination theory and draws upon the relationship between knowledge and industry to distinguish between basic knowledge search diversity and applied knowledge search diversity. It further investigates the distinct relationships between these two forms of knowledge search diversity and their respective contributions to breakthrough technological innovations. We develop research hypotheses and employ a negative binomial fixed-effects regression model to empirically test data from 280 Chinese new energy vehicle (NEV) manufacturers. The results demonstrate that: (1) knowledge search diversity exhibits an inverted U-shaped relationship with breakthrough technological innovation; (2) the inverted U-shaped relationship is steeper for basic knowledge search diversity compared to applied knowledge search diversity; (3) organizational proximity positively moderates the inverted U-shaped relationship between knowledge search diversity and breakthrough technological innovation. The findings provide practical insights into optimizing knowledge search strategies for technological convergence and accelerating the internalization of external resources to foster innovative transformation and upgrading.
- 26 June 2025
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 20 (6) , e0326736
Open AccessDiversity in knowledge search paths and breakthrough technological innovation: Evidence from new energy vehicle firms
Breakthrough technological innovations have heralded a transformative era within the automotive industry. Knowledge search diversity, as the key to acquiring the raw materials for knowledge recombination, enables firms to overcome resource constraints and accelerate the realization of breakthrough technological innovations. This paper integrates the theories of organizational search and knowledge recombination theory and draws upon the relationship between knowledge and industry to distinguish between basic knowledge search diversity and applied knowledge search diversity. It further investigates the distinct relationships between these two forms of knowledge search diversity and their respective contributions to breakthrough technological innovations. We develop research hypotheses and employ a negative binomial fixed-effects regression model to empirically test data from 280 Chinese new energy vehicle (NEV) manufacturers. The results demonstrate that: (1) knowledge search diversity exhibits an inverted U-shaped relationship with breakthrough technological innovation; (2) the inverted U-shaped relationship is steeper for basic knowledge search diversity compared to applied knowledge search diversity; (3) organizational proximity positively moderates the inverted U-shaped relationship between knowledge search diversity and breakthrough technological innovation. The findings provide practical insights into optimizing knowledge search strategies for technological convergence and accelerating the internalization of external resources to foster innovative transformation and upgrading.
- 28 April 2022
- Published by Elsevier in Applied Mathematics and Computation
- Vol. 428
Evaluating tacit knowledge diffusion with algebra matrix algorithm based social networks
- 28 April 2022
- Published by Elsevier in Applied Mathematics and Computation
- Vol. 428
Evaluating tacit knowledge diffusion with algebra matrix algorithm based social networks
- 1 November 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Annual International Conference of the IEEE Engineering in Medicine and Biology Society
- Vol. 2021, 500-503
Removing EOG Artifacts from the EEG signal of Methamphetamine Addicts
EEG can be used to characterize the electrical activity of the cerebral cortex, but it is also susceptible to interference. Compared with the other artifacts, Electrooculogram (EOG) artifacts have a greater impact on EEG processing and are more difficult to remove. Here, we mainly compared the regression and ICA algorithms both based on the EOG channels for the effect of removing EOG artifacts in the Stroop experiment of methamphetamine addicts. From the perspective of time domain and power spectral density, the ICA algorithm based on the EOG channels is more effective. However, the regression algorithm based on the EOG channels is less complex, more time-saving, and more suitable for real-time tasks.Clinical Relevance— For clinical purposes, this research has a certain reference value for selecting appropriate methods of removing EOG artifacts when processing the EEG of methamphetamine addicts.
- 1 November 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Annual International Conference of the IEEE Engineering in Medicine and Biology Society
- Vol. 2021, 500-503
Removing EOG Artifacts from the EEG signal of Methamphetamine Addicts
EEG can be used to characterize the electrical activity of the cerebral cortex, but it is also susceptible to interference. Compared with the other artifacts, Electrooculogram (EOG) artifacts have a greater impact on EEG processing and are more difficult to remove. Here, we mainly compared the regression and ICA algorithms both based on the EOG channels for the effect of removing EOG artifacts in the Stroop experiment of methamphetamine addicts. From the perspective of time domain and power spectral density, the ICA algorithm based on the EOG channels is more effective. However, the regression algorithm based on the EOG channels is less complex, more time-saving, and more suitable for real-time tasks.Clinical Relevance— For clinical purposes, this research has a certain reference value for selecting appropriate methods of removing EOG artifacts when processing the EEG of methamphetamine addicts.
- Open Access
- 26 April 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 68240-68252
Tracking Attention of Social Media Event by Hidden Markov Model–Cases from Sina Weibo
Sina Weibo has significantly impact on the information diffusion processes in many real-world social events. A large number of active users on Sina Weibo not only push the opinion diffusion, but also increase the influence abilities of events which conversely attracted much attentions of users to follow them. How to effectively track the event attention of users is one of the most important channels to get the public opinions. In order to predict the event attention more accurately, motivated by observations of social events' influence concerning with users and microblogs, we quantify the user popularity from the four dimensions: the user activity, the user behavior, the user authenticity and the user infection ability. And the non-collinearity of these four dimensions is tested to ensure the comprehensiveness and non-redundancy of the evaluation. Then, combining with the logic framework of Hidden Markov Model, we propose an algorithm to predict the Weibo event attention by using the user popularity. Meanwhile, in order to better detect the performance of the prediction algorithm, we integrate the static and dynamic information of microblog content to directly quantify the current Weibo event attention as a benchmark, and the performance of four prediction algorithms (including our algorithm) is tested with six real data sets which are chosen from the popular events in China from 2019 to 2020. Through comparison, we find that the user popularity can be used to predict the event attention, and the Hidden Markov Model prediction method by using the user popularity shows good prediction performance.
- 26 April 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 68240-68252
Open AccessTracking Attention of Social Media Event by Hidden Markov Model–Cases from Sina Weibo
Sina Weibo has significantly impact on the information diffusion processes in many real-world social events. A large number of active users on Sina Weibo not only push the opinion diffusion, but also increase the influence abilities of events which conversely attracted much attentions of users to follow them. How to effectively track the event attention of users is one of the most important channels to get the public opinions. In order to predict the event attention more accurately, motivated by observations of social events' influence concerning with users and microblogs, we quantify the user popularity from the four dimensions: the user activity, the user behavior, the user authenticity and the user infection ability. And the non-collinearity of these four dimensions is tested to ensure the comprehensiveness and non-redundancy of the evaluation. Then, combining with the logic framework of Hidden Markov Model, we propose an algorithm to predict the Weibo event attention by using the user popularity. Meanwhile, in order to better detect the performance of the prediction algorithm, we integrate the static and dynamic information of microblog content to directly quantify the current Weibo event attention as a benchmark, and the performance of four prediction algorithms (including our algorithm) is tested with six real data sets which are chosen from the popular events in China from 2019 to 2020. Through comparison, we find that the user popularity can be used to predict the event attention, and the Hidden Markov Model prediction method by using the user popularity shows good prediction performance.
- 1 January 2021
- Published by Elsevier in Knowledge-Based Systems
- Vol. 212, 106588
Understanding the game behavior with sentiment and unequal status in cooperation network
- 1 January 2021
- Published by Elsevier in Knowledge-Based Systems
- Vol. 212, 106588
Understanding the game behavior with sentiment and unequal status in cooperation network