Sai Gu
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
Sai Gu has a total of 61 co-authors. Here are the top co-authors by number of shared publications:
- Tomas Ramirez Reina
University of Seville
5 shared publications Arvind Kumar
Department of Mechanical Engineering, Indian Institute of Technology, Kanpur 208016, India
4 shared publications- Dekui Shen
Interdisciplinary Innovation Laboratory for Environment and Energy, School of Mechanical Engineering, Tianjin University of Commerce, Tianjin, China.
4 shared publications - Mahrukh Mahrukh
School of Intelligent Manufacturing and Control Engineering, Qilu Institute of Technology, Jinan, Shandong, P.R. China
4 shared publications - Laura Pastor-Pérez
Department of Inorganic Chemistry and Material Sciences Institute of Seville, University of Seville-CSIC, Seville 41092, Spain; Department of Chemical and Process Engineering, University of Surrey, Guildford GU2 7XH, United Kingdom
3 shared publications - Spyros Kamnis
Castolin Eutectic-Monitor Coatings Ltd
3 shared publications
Latest Publications
- 12 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. PP, 1-11
MSDiff: Multi-Strategy Diffusion Model for Social Behavior Modeling and Recommendation
Understanding and simulating user behavior in social networks is fundamental to computational social intelligence. Traditional recommendation methods rely on discriminative models that predict user preferences but fail to capture the generative processes underlying social influence propagation and behavior formation. In this article, we propose multi-strategy diffusion (MSDiff), a MSDiff model that leverages generative AI to model, simulate, and enhance social user behavior for recommendation. Our approach introduces three key innovations: 1) a selective diffusion mechanism that specifically activates dormant users by simulating latent social influence propagation, effectively modeling behavior emergence in inactive populations; 2) a differential generative strategy that treats user-user (U-U) and user-item (U-I) interactions distinctly—using U-U diffusion to simulate social homophily dynamics and U-I diffusion to generate contrastive behavioral samples; and 3) a collaborative graph convolution layer that integrates multirelational signals to model complex social dynamics. By framing social recommendation as a generative modeling problem, MSDiff not only improves recommendation accuracy but also provides interpretable simulations of how user preferences form and propagate through social networks. Extensive experiments on three real-world datasets demonstrate that MSDiff significantly outperforms state-of-the-art baselines, validating its effectiveness in both behavior simulation and recommendation tasks.
- 12 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. PP, 1-11
MSDiff: Multi-Strategy Diffusion Model for Social Behavior Modeling and Recommendation
Understanding and simulating user behavior in social networks is fundamental to computational social intelligence. Traditional recommendation methods rely on discriminative models that predict user preferences but fail to capture the generative processes underlying social influence propagation and behavior formation. In this article, we propose multi-strategy diffusion (MSDiff), a MSDiff model that leverages generative AI to model, simulate, and enhance social user behavior for recommendation. Our approach introduces three key innovations: 1) a selective diffusion mechanism that specifically activates dormant users by simulating latent social influence propagation, effectively modeling behavior emergence in inactive populations; 2) a differential generative strategy that treats user-user (U-U) and user-item (U-I) interactions distinctly—using U-U diffusion to simulate social homophily dynamics and U-I diffusion to generate contrastive behavioral samples; and 3) a collaborative graph convolution layer that integrates multirelational signals to model complex social dynamics. By framing social recommendation as a generative modeling problem, MSDiff not only improves recommendation accuracy but also provides interpretable simulations of how user preferences form and propagate through social networks. Extensive experiments on three real-world datasets demonstrate that MSDiff significantly outperforms state-of-the-art baselines, validating its effectiveness in both behavior simulation and recommendation tasks.
- Open Access
- 12 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 1640-1644
On Improving PPG-Based Sleep Staging: A Pilot Study
- 12 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 1640-1644
Open AccessOn Improving PPG-Based Sleep Staging: A Pilot Study
- 19 November 2020
- Published by American Chemical Society (ACS) in Industrial & Engineering Chemistry Research
- Vol. 59 (51) , 22017-22039
State-of-the-Art on the Preparation, Modification, and Application of Biomass-Derived Carbon Quantum Dots
- 19 November 2020
- Published by American Chemical Society (ACS) in Industrial & Engineering Chemistry Research
- Vol. 59 (51) , 22017-22039
State-of-the-Art on the Preparation, Modification, and Application of Biomass-Derived Carbon Quantum Dots
- 12 October 2020
- Published by American Chemical Society (ACS) in ACS Sustainable Chemistry & Engineering
- Vol. 8 (43) , 16217-16228
Nonprecious Metal/Bimetallic Catalytic Hydrogenolysis of Lignin in a Mixed-Solvent System
- 12 October 2020
- Published by American Chemical Society (ACS) in ACS Sustainable Chemistry & Engineering
- Vol. 8 (43) , 16217-16228
Nonprecious Metal/Bimetallic Catalytic Hydrogenolysis of Lignin in a Mixed-Solvent System
- 16 September 2020
- Published by American Chemical Society (ACS) in ACS Omega
- Vol. 5 (38) , 24684-24692
Experimental Investigation of the Impact of CO, C2H6, and H2 on the Explosion Characteristics of CH4
Gas explosions are destructive disasters in coal mines. Coal mine gas is a multi-component gas mixture, with methane (CH4) being the dominant constituent. Understanding the process and mechanism of mine gas explosions is of critical importance to the safety of mining operations. In this work, three flammable gases (CO, C2H6, and H2) which are commonly present in coal mines were selected to explore how they affect a methane explosion. The explosion characteristics of the flammable gases were investigated in a 20 L spherical closed vessel. Experiments on binary- (CH4/CO, CH4/C2H6, and CH4/H2) and multicomponent (CH4/CO/C2H6/H2) mixtures indicated that the explosion of such mixtures is more dangerous and destructive than that of methane alone in air, as measured by the explosion pressure. Furthermore, a self-promoting microcirculation reaction network is proposed to help analyze the chemical reactions involved in the multicomponent (CH4/CO/C2H6/H2) gas explosion. This work will contribute to a better understanding of the explosion mechanism of gas mixtures in coal mines and provide a useful reference for determining the safety limits in practice.
- 16 September 2020
- Published by American Chemical Society (ACS) in ACS Omega
- Vol. 5 (38) , 24684-24692
Experimental Investigation of the Impact of CO, C2H6, and H2 on the Explosion Characteristics of CH4
Gas explosions are destructive disasters in coal mines. Coal mine gas is a multi-component gas mixture, with methane (CH4) being the dominant constituent. Understanding the process and mechanism of mine gas explosions is of critical importance to the safety of mining operations. In this work, three flammable gases (CO, C2H6, and H2) which are commonly present in coal mines were selected to explore how they affect a methane explosion. The explosion characteristics of the flammable gases were investigated in a 20 L spherical closed vessel. Experiments on binary- (CH4/CO, CH4/C2H6, and CH4/H2) and multicomponent (CH4/CO/C2H6/H2) mixtures indicated that the explosion of such mixtures is more dangerous and destructive than that of methane alone in air, as measured by the explosion pressure. Furthermore, a self-promoting microcirculation reaction network is proposed to help analyze the chemical reactions involved in the multicomponent (CH4/CO/C2H6/H2) gas explosion. This work will contribute to a better understanding of the explosion mechanism of gas mixtures in coal mines and provide a useful reference for determining the safety limits in practice.
- Open Access
Transition Metal Carbides (TMCs) Catalysts for Gas Phase CO2 Upgrading Reactions: A Comprehensive Overview
Increasing demand for CO2 utilization reactions and the stable character of CO2 have motivated interest in developing highly active, selective and stable catalysts. Precious metal catalysts have been studied extensively due to their high activities, but their implementation for industrial applications is hindered due to their elevated cost. Among the materials which have comparatively low prices, transition metal carbides (TMCs) are deemed to display catalytic properties similar to Pt-group metals (Ru, Rh, Pd, Ir, Pt) in several reactions such as hydrogenation and dehydrogenation processes. In addition, they are excellent substrates to disperse metallic particles. Hence, the unique properties of TMCs make them ideal substitutes for precious metals resulting in promising catalysts for CO2 utilization reactions. This work aims to provide a comprehensive overview of recent advances on TMCs catalysts towards gas phase CO2 utilization processes, such as CO2 methanation, reverse water gas shift (rWGS) and dry reforming of methane (DRM). We have carefully analyzed synthesis procedures, performances and limitations of different TMCs catalysts. Insights on material characteristics such as crystal structure and surface chemistry and their connection with the catalytic activity are also critically reviewed.
Open AccessTransition Metal Carbides (TMCs) Catalysts for Gas Phase CO2 Upgrading Reactions: A Comprehensive Overview
Increasing demand for CO2 utilization reactions and the stable character of CO2 have motivated interest in developing highly active, selective and stable catalysts. Precious metal catalysts have been studied extensively due to their high activities, but their implementation for industrial applications is hindered due to their elevated cost. Among the materials which have comparatively low prices, transition metal carbides (TMCs) are deemed to display catalytic properties similar to Pt-group metals (Ru, Rh, Pd, Ir, Pt) in several reactions such as hydrogenation and dehydrogenation processes. In addition, they are excellent substrates to disperse metallic particles. Hence, the unique properties of TMCs make them ideal substitutes for precious metals resulting in promising catalysts for CO2 utilization reactions. This work aims to provide a comprehensive overview of recent advances on TMCs catalysts towards gas phase CO2 utilization processes, such as CO2 methanation, reverse water gas shift (rWGS) and dry reforming of methane (DRM). We have carefully analyzed synthesis procedures, performances and limitations of different TMCs catalysts. Insights on material characteristics such as crystal structure and surface chemistry and their connection with the catalytic activity are also critically reviewed.
- Open Access
- 8 June 2020
- Published by American Chemical Society (ACS) in Industrial & Engineering Chemistry Research
- Vol. 59 (27) , 12525-12535
Pattern Matching and Active Simulation Method for Process Fault Diagnosis
- 8 June 2020
- Published by American Chemical Society (ACS) in Industrial & Engineering Chemistry Research
- Vol. 59 (27) , 12525-12535
Open AccessPattern Matching and Active Simulation Method for Process Fault Diagnosis
- 1 February 2020
- Published by Elsevier in Journal of the Energy Institute
- Vol. 93 (1) , 281-291
Co-pyrolysis of lignin and polyethylene with the addition of transition metals - Part I: Thermal behavior and kinetics analysis
- 1 February 2020
- Published by Elsevier in Journal of the Energy Institute
- Vol. 93 (1) , 281-291
Co-pyrolysis of lignin and polyethylene with the addition of transition metals - Part I: Thermal behavior and kinetics analysis
- Open Access
Catalytic Conversion of Palm Oil to Bio-Hydrogenated Diesel over Novel N-Doped Activated Carbon Supported Pt Nanoparticles
Bio-hydrogenated diesel (BHD), derived from vegetable oil via hydrotreating technology, is a promising alternative transportation fuel to replace nonsustainable petroleum diesel. In this work, a novel Pt-based catalyst supported on N-doped activated carbon prepared from polypyrrole as the nitrogen source (Pt/N-AC) was developed and applied in the palm oil deoxygenation process to produce BHD in a fixed bed reactor system. High conversion rates of triglycerides (conversion of TG > 90%) and high deoxygenation percentage (DeCOx% = 76% and HDO% = 7%) were obtained for the palm oil deoxygenation over Pt/N-AC catalyst at optimised reaction conditions: T = 300 °C, 30 bar of H2, and LHSV = 1.5 h−1. In addition to the excellent performance, the Pt/N-AC catalyst is highly stable in the deoxygenation reaction, as confirmed by the XRD and TEM analyses of the spent sample. The incorporation of N atoms in the carbon structure alters the electronic density of the catalyst, favouring the interaction with electrophilic groups such as carbonyls, and thus boosting the DeCOx route over the HDO pathway. Overall, this work showcases a promising route to produce added value bio-fuels from bio-compounds using advanced N-doped catalysts.
Open AccessCatalytic Conversion of Palm Oil to Bio-Hydrogenated Diesel over Novel N-Doped Activated Carbon Supported Pt Nanoparticles
Bio-hydrogenated diesel (BHD), derived from vegetable oil via hydrotreating technology, is a promising alternative transportation fuel to replace nonsustainable petroleum diesel. In this work, a novel Pt-based catalyst supported on N-doped activated carbon prepared from polypyrrole as the nitrogen source (Pt/N-AC) was developed and applied in the palm oil deoxygenation process to produce BHD in a fixed bed reactor system. High conversion rates of triglycerides (conversion of TG > 90%) and high deoxygenation percentage (DeCOx% = 76% and HDO% = 7%) were obtained for the palm oil deoxygenation over Pt/N-AC catalyst at optimised reaction conditions: T = 300 °C, 30 bar of H2, and LHSV = 1.5 h−1. In addition to the excellent performance, the Pt/N-AC catalyst is highly stable in the deoxygenation reaction, as confirmed by the XRD and TEM analyses of the spent sample. The incorporation of N atoms in the carbon structure alters the electronic density of the catalyst, favouring the interaction with electrophilic groups such as carbonyls, and thus boosting the DeCOx route over the HDO pathway. Overall, this work showcases a promising route to produce added value bio-fuels from bio-compounds using advanced N-doped catalysts.