Effect Of Embedding Research Articles

2024 Presidential Election Sentiment Analysis in News Media Using Support Vector Machine

The 2024 presidential election is an event for all Indonesian people to determine their best leader. The presidential and vice presidential candidates are also competing to give their best efforts so that they can be elected as President and Vice President. The news media also provide news related to the 2024 presidential election with various titles that can interest their readers. Not infrequently the titles raised contain words that have sentiments, both positive and negative. In order to facilitate the analysis of the sentiments of these news titles, it is necessary to build a system that can detect the sentiments of these titles. In this study, we built a sentiment analysis system using the Support Vector Machine (SVM) method on news headline data obtained from online news media to detect whether news headlines contain positive or negative sentiments. For feature exctraction we compare the effect of FastText word embedding with TF-IDF for feature extraction. In the SVM method, several experiments were carried out on the attributes of C, kernel, gamma, and the ratio of the test data. The results obtained are a FastText can outperform TF-IDF for feature extraction. Also, combination of Kernel, C, and gamma values that give the best accuracy score of rbf, 1, and auto respectively at a test data ratio of 90:10, with an accuracy score of 99%.

A Study on the Performance of Recurrent Neural Network based Models in Maithili Part of Speech Tagging

This article presents our effort in developing a Maithili Part of Speech (POS) tagger. Substantial effort has been devoted to developing POS taggers in several Indian languages, including Hindi, Bengali, Tamil, Telugu, Kannada, Punjabi, and Marathi; but Maithili did not achieve much attention from the research community. Maithili is one of the official languages of India, with around 50 million native speakers. So, we worked on developing a POS tagger in Maithili. For the development, we use a manually annotated in-house Maithili corpus containing 56,126 tokens. The tagset contains 27 tags. We train a conditional random fields (CRF) classifier to prepare a baseline system that achieves an accuracy of 82.67%. Then, we employ several recurrent neural networks (RNN)-based models, including Long-short Term Memory (LSTM), Gated Recurrent Unit (GRU), LSTM with a CRF layer (LSTM-CRF), and GRU with a CRF layer (GRU-CRF) and perform a comparative study. We also study the effect of both word embedding and character embedding in the task. The highest accuracy of the system is 91.53%.

Multilayer self‐attention residual network for code search

SummarySoftware developers usually search existing code snippets in open source code repositories to modify and reuse them. Therefore, how to get the right code snippet from the open‐source code repository quickly and accurately is the focus of current software development research. Nowadays, code search is one of the solutions. To improve the accuracy of source code feature information representation and the accuracy of code search. A multilayer self‐ attention residual network‐based code search model (MSARN‐CS) is proposed in this paper. In the MSARN‐CS model, not only the weight of each word in the code sequence unit is considered but also the effect of embedding between code sequence units is calculated. In addition, an optimization model of residuals is introduced to compensate for the loss of information in the code sequences during the model training. To verify the search effectiveness of the MSARN‐CS model, three other baseline models are compared on the basis of extensive source code data. The experimental results show that the MSARN‐CS model has better search results compared with the baseline model. For parameter Recall@1, the experimental result of MSARN‐CS model was 9.547, which as 100.90%, 73.87%, 60.37%, and 2.55% better compared to CODEnn, CRLCS, SAN‐CS‐ and SAN‐CS, respectively. For the parameter Recall@5, the results improved by 26.67%, 36.23%, 36.21%, and 1.63%, respectively, and for the parameter Recall@10, the results improved by 13.92%, 25.70%, 20.78%, and 2.23%, respectively. For the parameter mean reciprocal rank, the results improved by 52.89%, 76.17%, 63.38%, and 3.88%, respectively. For the parameter normalized discounted cumulative gain, the results improved by 54.22%, 60.55%, 50.28%, and 3.30%, respectively. The MSARN‐CS model proposed in the paper can effectively improve the accuracy of code search and enhance the programming efficiency of developers.

Poly(amidoamine) dendrimer-induced 3D crosslinked network constructed on polyphenylene sulfide nonwoven as a battery separator: Effect of generation number on cell performance

High power lithium-ion batteries put forward challenges for separator in terms of meeting high current density, safety as well as superior cycling stability. In this study, Poly(amidoamine) (PAMAM) dendrimer-introduced three-dimensional (3D) crosslinked coating layer is built on polyphenylene sulfide (PPS) nonwoven substrate to address above issues and the generation number effect are discussed thoroughly. Due to unique advantages of PAMAM dendrimer including its hydrophilicity to electrolyte, embedding effect of internal cavities, and spherical steric structure, the porosity, ionic conductivity, interfacial compatible as well as lithium-ion transference number are improved for the crosslinked composite separator especially at higher generation number, which endow battery with superior discharge capacity of 162.5 mAh g−1 at 0.2 C and outstanding C-rate capabilities. Moreover, through the adjustment of generation number, less crosslinking junctions but more crosslinking points at one junction can be achieved in PAMAM-induced 3D crosslinked structures, thus leading to the improvement of all mechanical properties concurrently. This behavior can shift the oxidative decomposition potential to 5.36 V and give superior cell performance retention (98.6%) even after 100 cycles for battery using the fourth generation PAMAM. Therefore, this study will provide a new route for separator when pursuing higher rate performance and cyclic stability.

Robust and blind watermarking algorithm for chaotic system encryption based on discrete cosine transform-singular value decomposition

Through comparison with most frequency-domain digital watermarking approaches, the spread transform dither modulation (STDM) method exhibits a good embedding effect and is capable of meeting the requirements of blind extraction. However, the rotating attack is difficult to counter, and the cropping attack is seriously impacted by location factors. In addition, there exist numerous algorithms concentrate solely on the watermarking process, ignoring the image’s security. In these views, our work proposes a two-dimensional hyperchaotic system, the so-named Chebyshev ICMIC cascade map with the Knuth–Durstenfeld method, aiming to enhance the security of digital watermarking systems. The scrambled pixels improve the anticropping attack effect through breaking the location correlation of watermark information. For the purpose of magnifying resistance to rotation attack impact, the radial harmonic Fourier moments subtraction coefficient matrix distribution expectation is appropriately adapted to carry out the rotation correction. The performed experiments reveal that the suggested improved STDM watermarking algorithm demonstrates better robustness and invisibility, resisting more attacks than the existing methodologies. Finally, the feasibility of the proposed methods is proved based on the simulation results.

Detecting Vocal Fatigue with Neural Embeddings.

Vocal fatigue refers to the feeling of tiredness and weakness of voice due to extended utilization. This paper investigates the effectiveness of neural embeddings for the detection of vocal fatigue. We compare x-vectors, ECAPA-TDNN, and wav2vec 2.0 embeddings on a corpus of academic spoken English. Low-dimensional mappings of the data reveal that neural embeddings capture information about the change in vocal characteristics of a speaker during prolonged voice usage. We show that vocal fatigue can be reliably predicted using all three types of neural embeddings after 40 minutes of continuous speaking when temporal smoothing and normalization are applied to the extracted embeddings. We employ support vector machines for classification and achieve accuracy scores of 81% using x-vectors, 85% using ECAPA-TDNN embeddings, and 82% using wav2vec 2.0 embeddings as input features. We obtain an accuracy score of 76%, when the trained system is applied to a different speaker and recording environment without any adaptation.

Learning Heterogeneous Graph Embedding with Metapath-Based Aggregation for Link Prediction

Along with the growth of graph neural networks (GNNs), many researchers have adopted metapath-based GNNs to handle complex heterogeneous graph embedding. The conventional definition of a metapath only distinguishes whether there is a connection between nodes in the network schema, where the type of edge is ignored. This leads to inaccurate node representation and subsequently results in suboptimal prediction performance. In heterogeneous graphs, a node can be connected by multiple types of edges. In fact, each type of edge represents one kind of scene. The intuition is that if the embedding of nodes is trained under different scenes, the complete representation of nodes can be obtained by organically combining them. In this paper, we propose a novel definition of a metapath whereby the edge type, i.e., the relation between nodes, is integrated into it. A heterogeneous graph can be considered as the compound of multiple relation subgraphs from the view of a novel metapath. In different subgraphs, the embeddings of a node are separately trained by encoding and aggregating the neighbors of the intrapaths, which are the instance levels of a novel metapath. Then, the final embedding of the node is obtained by the use of the attention mechanism which aggregates nodes from the interpaths, which is the semantic level of the novel metapaths. Link prediction is a downstream task by which to evaluate the effectiveness of the learned embeddings. We conduct extensive experiments on three real-world heterogeneous graph datasets for link prediction. The empirical results show that the proposed model outperforms the state-of-the-art baselines; in particular, when comparing it to the best baseline, the F1 metric is increased by 10.35% over an Alibaba dataset.

Does Global Value Chain Embedment Contribute to Environmental Pollution in Emerging Economies?

Scientifically judging the environmental effects of embedding in the global value chain (GVC) has important theoretical significance and practical value, especially for emerging economies, but there are few studies on the relationship between GVC embedment and environmental pollution in emerging economies. Based on the classical theory of Grossman, this study analyzed the mechanism framework of GVC embedment on environmental pollution and then used the panel data of 16 emerging economies from 1998 to 2019 for empirical tests. After using the FGLS regression model and a series of empirical tests, the main conclusions are as follows. (1) The impact of GVC embedment on environmental pollution in emerging economies includes five effects. The scale effect brings about the expansion of production activities and increases the emission of environmental pollutants. The technological effect refers to improvements in the level of clean technology, which are brought about by the acquisition and acceptance of “technology spillover” in the value chain. The structural effect is the reduction in pollutant emissions and improvement in the environment through economic upgrading. The income effect is a “demand push” to improve the environment and reduce pollution triggered by an increase in national income. The lock-in effect means that the leaders of the value chain hinder emerging economies from achieving high-level economic upgrades, intending to lock them into the links with relatively high pollution. These effects are not static but are interlinked and affect each other. (2) The results of the empirical study show that embedding in the GVC initially has a negative impact on the environment, but with the deepening of GVC embedment, environmental pollution will decrease; that is, it shows an inverted U-shaped relationship. (3) The density of energy use, the share of fossil fuels, and the per capita income of emerging economies are positively correlated with environmental pollution, while population density and manufacturing share are negatively correlated. This study also provides suggestions on how to improve the environment in the process of GVC embedment in the future.

ReLOAD: Using Reinforcement Learning to Optimize Asymmetric Distortion for Additive Steganography

Recently, the success of non-additive steganography has demonstrated that asymmetric distortion can remarkably improve security performance compared with symmetric cost functions. However, most of current existing additive steganographic methods are still based on symmetric distortion. In this paper, for the first time we optimize asymmetric distortion for additive steganography and propose an A3C(Asynchronous Advantage Actor-Critic) based steganographic framework, called ReLOAD. ReLOAD is composed of an actor and a critic, where the former guides action selection for pixel-wise distortion modulation, and the latter evaluates the performance of modulated distortion. Meanwhile, a reward function that considers embedding effects is proposed to unify the goal of steganography and reinforcement learning, so that the minimization of embedding effects can be achieved by learning secure policy to maximize total rewards. Statistical analysis shows that compared with non-additive steganography, ReLOAD achieves lower change rates and makes embedding traces more consistent with cover image textures. Comprehensive experiments conducted on both hand-crafted feature-based and deep learning-based steganalyzers show that ReLOAD significantly promotes the state-of-the-art security performance of current additive methods and even outperforms non-additive steganography when the modification distribution gets sparser.

Multi-Channel HEVC Steganography by Minimizing IPM Steganographic Distortions

Calibration is a common method for steganalysis, and Intra Prediction Mode (IPM) shift is a typical phenomenon used in calibration to detect video steganography. The current HEVC steganography lacks resistance to steganalysis based on this phenomenon because the new technology of HEVC introduces steganographic distortion in addition to providing more potential steganographic space. In this paper, an HEVC steganographic algorithm that resists IPM shift is proposed. First, we introduce the IPM shift in HEVC, and the previous H.264 steganalytic IPM shift feature is modeled and improved. By analyzing the HEVC encoding process, we found that modifying large-size blocks has a more significant impact on compression efficiency, while small ones are more sensitive to IPM optimality. Therefore, we perform the embedding channel division based on the block size and design the distortion function separately. In addition, we discover a unique IPM transition probability distribution in HEVC. According to our analysis, this unique distribution arises due to HEVC's MPM rules and the regularity of IPM direction. Modifying IPM in HEVC will change such distribution, thus, a mapping rule is designed based on this distribution to achieve a better embedding effect. Experimental results show that the channel division and proposed distortion function can effectively improve the overall performance. The proposed steganography outperforms the state-of-the-art steganography in resisting steganalysis, bitrate controlling, and visual quality.

Cross Modal Video Representations for Weakly Supervised Active Speaker Localization

An objective understanding of media depictions, such as inclusive portrayals of how much someone is heard and seen on screen such as in film and television, requires the machines to discern automatically who, when, how, and where someone is talking, and not. Speaker activity can be automatically discerned from the rich multimodal information present in the media content. This is however a challenging problem due to the vast variety and contextual variability in the media content, and the lack of labeled data. In this work, we present a cross-modal neural network for learning visual representations, which have implicit information pertaining to the spatial location of a speaker in the visual frames. Avoiding the need for manual annotations for active speakers in visual frames, acquiring of which is very expensive, we present a weakly supervised system for the task of localizing active speakers in movie content. We use the learned cross-modal visual representations, and provide weak supervision from movie subtitles acting as a proxy for voice activity, thus requiring no manual annotations. We evaluate the performance of the proposed system on the AVA active speaker dataset and demonstrate the effectiveness of the cross-modal embeddings for localizing active speakers in comparison to fully supervised systems. We also demonstrate state-of-the-art performance for the task of voice activity detection in an audio-visual framework, especially when speech is accompanied by noise and music.

Preload loss of CFRP bolted joint without rotation under transverse and axial loading

There are a large number of CFRP-steel bolted joints with jet nuts subjected to transverse and axial load in aircraft structures, which have passed standard testing of locking characteristics of fasteners under transverse loading conditions. This fastener has a great preload loss and no rotation loosening occurs. Based on the observation, we designed a biaxial loading experiment for CFRP bolted joint, and measured the preload and loosening angle under different initial preload. The biaxial research suggests that compared with the uniaxial loading condition, the preload loss is larger, the rotation is minimal and only appears in the early stage of test, and there are obvious indentations on the surface of CFRP panel. Embedment as an important factor leads to the preload loss due to the weak ability of CFRP panel suffering out of plane load. In order to study the effect of embedment, the relationship between stress S22 and indentation depth under different preload, the relationship between biaxial load amplitude and embedding depth, and the ratio of the preload loss caused by embedding to the total preload loss are revealed by finite element analysis.

Comparing acupoint catgut embedding and acupuncture therapies for simple obesity: A systematic review and meta-analysis

This systematic review aimed to assess the effectiveness and safety of acupoint catgut embedding (ACE), which is widely used in simple obese patients. Nine electronic databases and 2 trial registries were searched from inception to September 2021 without language limitations. All randomized controlled trials involving ACE therapies for simple obesity were included. Assessment of Cochrane's risk of bias and meta-analysis, as well as GRADE evaluation, were also performed. A total of 73 randomized clinical trials involving 5872 participants were included. The overall risk of bias was high or unclear. ACE showed higher efficacy in total effective rate, body mass index, and body weight than manual acupuncture. In addition, ACE is more beneficial in total effective rate than electroacupuncture (EA). Moreover, combination therapy of ACE with EA is more effective in total body mass index and body weight than EA alone. Despite some potential improvement, the evidence regarding the effectiveness and efficacy of ACE for simple obesity is inconclusive due to the poor quality of evidence. Further well-designed randomized controlled trials are needed to confirm the effectiveness of ACE for simple obesity.

Time evolution of elastic stiffness coefficients for a jack-up spudcan in clay during operation

Current design practice of spudcans in clay generally ignores the influence of installation and subsequent operation on the spudcan elastic stiffness. In reality, soil consolidation around the post-installation spudcan can cause the elastic stiffness to evolve with time, thus affecting the safe and stable operation of jack-up platforms. This paper reports a numerical study dedicated to this specific subject. The spudcan installation and the time evolution of elastic stiffness coefficients during operation are continuously simulated using an effective stress large deformation finite element technique. The effects of soil, loading, embedment and geometric conditions were evaluated and interpreted. It was found that the non-uniform growth of the shear modulus caused by soil consolidation during operation results in the time evolution of the elastic stiffness and makes the deformation mechanism asymmetric. The overlap degree between the involved zone of the deformation mechanism and the growth zone of the shear modulus (or the so-called averaging effect) affects the gradient of time evolution. Based on the parametric study, a generalized framework is present for predicting the elastic stiffness coefficients of spudcans for different scenarios and consolidation times, thus providing a complete view of the time evolution of elastic stiffness for the offshore in-site assessment.

Systematic review and Meta-analysis of acupuncture and acupoint catgut embedding for the treatment of abdominal obesity.

To systematically review and analyze the effect of acupuncture and acupoint catgut embedding in treatment of abdominal obesity to provide a more reasonable clinical treatment regimen. Ten databases were searched as of August 2022: the English databases PubMed, Embase, Cochrane Library, Web of Science, Wiley, and Scopus and the Chinese databases China National Knowledge Infrastructure Database, China Science and Technology Journal Database, Wanfang, and SinoMed/Chinese Biomedical Literature Database. Randomized controlled trials (RCTs) of acupuncture and acupoint catgut embedding as the main interventions to treat abdominal obesity were extracted. The investigators imported the citations into EndNote version X9.1 for deduplication, screening, extraction, and integration. The risk of bias in the included RCTs was assessed according to the Cochrane Handbook. RevMan 5.4 software was used to conduct a Meta-analysis of RCTs that met the inclusion criteria. Thirteen RCTs (1069 patients) were included in this study, and the data of eleven RCTs (966 patients) were include in the Meta-analysis. The results showed that acupoint catgut embedding can significantly change the weight and waist circumference of patients with abdominal obesity when compared to sham acupuncture or no treatment [mean difference () = 2.32, 95% confidence interval () (1.88, 2.76), < 0.000 01], [ = 3.47, 95% (1.99, 4.94), < 0.000 01]. The change in hip circumference after acupuncture was also significant [ = 0.89, 95% (0.12, 1.66), = 0.02]. This study found that acupuncture and acupoint catgut embedding can effectively treat abdominal obesity, therefore, these interventions can be used as clinical supplements and alternative therapies. The diagnostic criteria of the existing studies and the intervention measures of the control group are not unified. It will be necessary to improve the clinical study protocols and expand the sample size to further validate the reliability of the results obtained of this study.

Acupoint catgut embedding versus acupuncture for simple obesity: a systematic review and Meta-analysis of randomized controlled trials.

To evaluate the clinical effectiveness and safety of acupoint catgut embedding and acupuncture on simple obesity by Meta-analysis. Studies on clinical randomized controlled trials of acupoint catgut embedding for simple obesity which were published from January 2015 to November 2020 were searched in Cochrane Central Register of Control Trials (Central), PubMed, China Science and Technology Journal Database, China National Knowledge Infrastructure Database and Wanfang data-bases. And those that met the inclusion criteria were screened. RevMan5.3 was used for Meta-analysis. The "Risk of Bias" tool was used to evaluate the quality of included studies. R studio software was used for the measurement of publication bias. A total of 33 studies were included for Meta-analysis, including 2685 patients with simple obesity. Meta-analysis results showed the comparison of effectiveness rate was relative risk () = 1.12, 95%(1.08, 1.16), body mass index (BMI) was mean difference () = -1.12, 95% (-2.09, -0.14), waist circumference was = -2.14, 95% (-4.22, -0.06), and body mass was = -2.36, 95% (-3.99, -0.73). On the basis of diet and exercise intervention, the effectiveness rate [ = 1.12, 95% (1.05, 1.19)], BMI [ = -0.88, 95% (-1.35, -0.40)], waist circumference [ = -1.10, 95%(-4.27, 2.07)], and body mass [ = -0.68, 95%(-2.90, 1.54)]. The risk of bias of included literatures was low. Acupoint catgut embedding therapy was slightly better than acupuncture therapy in most of the outcomes. Moreover, the treatment frequency of acupoint catgut embedding is less with larger stimulation intensity, which is more conducive to clinical promotion.

Preparation and characterization of high embedding efficiency epigallocatechin-3-gallate glycosylated nanocomposites

Glycosylated protein nano encapsulation was an efficient encapsulation technology, but its embedding rate for EGCG was not high, and the research on the embedding mechanism was relatively weak. Based on this, this study compared the embedding effect of glycosylated peanut globulin and glycosylated casein on EGCG. The embedding mechanism of EGCG with glycosylated protein was discussed by ultraviolet, fluorescence, infrared and fluorescence microscopy. Results revealed that the highest encapsulation efficiency of EGCG was 93.89 ± 1.11%. The neutral pH value and 0.3 mg/mL EGCG addition amount were suitable for EGCG glycosylated nanocomposites. The hydrogen bond between EGCG hydroxyl group and tyrosine and tryptophan of glycosylated protein is mainly non covalent. The encapsulation effect of EGCG glycosylated nanocomposites could be quenched by changing the polar environment and spatial structure of the group. The fluorescence characteristic and dispersibility of EGCG glycosylated peanut globin were higher than EGCG glycosylated casein. This study might provide a theoretical basis for EGCG microencapsulation technology and EGCG application in tea beverage and liquid tea food systems.

Estimating farmers’ willingness to pay for photovoltaic industry to improve agricultural green resources and environment

Based on the consideration of the level of the regional radiation effect of the photovoltaic industry, four main regions were selected as case sites in Pingluo County, Ningxia Province, for this research. The study used the double-bound dichotomous CVM model and C-optimal design to revise the bidding value of the payment scheme that reduced the WTP range error. Five independent sub-sample questionnaires, including seven bidding value payment schemes, along with four internal range tests and six external range tests, were designed. The results showed that the comparison of the WTP for the improvement of agricultural resources and the environment among different independent sub-samples in the same region passed the external quantitative scope test, indicating that the questionnaire design, survey implementation, and WTP estimation results of the double-bound dichotomous CVM in this study were reliable. Saturation degree, diminishing marginal utility, and substitution effect are the main influencing factors of low marginal WTP added value and embedding effect. When the quantity range changes, the utility of respondents tends to saturate and marginal utility rapidly decreases. Influenced by the aforementioned factors, the sum of the independent estimates of the respondents on the WTP for the improvement of agricultural green resources and environment of the photovoltaic industry in the four research areas is 3.04 times the average value of the overall estimates of the four areas. In order to obtain and maintain the local photovoltaic industry for agricultural production resources and environmental improvement, the average WTP of each respondent was 99.80 yuan per year.

Mechanical Response Analysis of Asphalt Pavement Structure with Embedded Sensor

Long-term and real-time monitoring of asphalt pavement can be carried out by using embedded sensors to perceive and predict structural damage during pavement operation period, so as to avoid sustained development of damage. However, the influence of embedded sensors on the mechanical properties of asphalt pavement structure and the structural optimization of sensing elements needs to be further studied. Based on the finite element numerical simulation method, static load model and three-point bending test mode were conducted with three “pavement-sensor” coupling model without sensor, with embedded I-shape sensor, with embedded corrugated-shape sensor. Three simulated conditions were studied comparatively of the sensing element embedding effect on the mechanical response of asphalt pavement structure. Results show that the sensing elements embedded with the two structures have a certain influence on the stress and strain field of asphalt concrete. Within the range of 60–100 mm the asphalt mixture is in a state of tension; the stress values increase with depth and show a maximum tensile stress state at the bottom of the beam. In the compression zone, the strain of the I-shape sensing element embedded is closer to that of the strain without the sensing element embedded. Along the axis of the two sensing elements, the axial strain of the I-shape sensing element is smoother and uniform, which ensures the deformation coordination in the road state. The optimal length L of the sensing element is 14 cm, the diameter φ of the sensor is 10 mm, and the I-beam length GL is 10 cm.

Impact of global value chain embedding on industrial environmental performance: An empirical study based on the countries along the “Belt and Road”

Global value chain (GVC) embedding is a “Double-edged sword”. While the countries along the “Belt and Road” benefit from the dividends brought by GVC embedding, the environmental pollution caused by industrial production becomes increasingly severe due to the lack of core technology and weak research and development (R&amp;amp;D) capability. Environmental performance can comprehensively reflect the cleaner production level of the industry considering the interaction of various input factors. With the deepening of GVC embedding in countries along the “Belt and Road”, it is essential to improve industrial environmental performance from the perspective of GVC embedding. Based on the industrial sector data from 15 countries along the “Belt and Road” during the period from 2007 to 2020, the impacts of GVC embedding on industrial environmental performance are empirically examined in this study using the feasible generalized least squares (FGLS) and panel threshold model. The findings of this study could shed light on industrial pollution emission reduction measures through GVC embedding for the “Belt and Road” countries. Specifically, the results manifest that: 1) The increase of GVC forward embedding promotes industrial environmental performance, while the increase of GVC backward embedding has an inhibitory effect. 2) The impact of GVC position on industrial environmental performance shows a evident threshold effect with respect to the intensity of environmental regulation. In other words, when a country’s environmental regulation intensity rises continuously and crosses the threshold value, the upgrade of the GVC position will promote industrial environmental performance. 3) The effect of GVC embedding on industrial environmental performance is heterogeneous. From the perspective of cross-border times of added value, GVC backward simple embedding inhibits industrial environmental performance, while GVC backward complex embedding is promotive. From the perspective of time heterogeneity, the implementation of the “Belt and Road” Initiative has changed the impact of GVC backward embedding on environmental performance from inhibition to promotion. In the end, this paper provides policy implications for countries along the “Belt and Road” to establish a green and low-carbon circular development system and achieve the goal of industrial clean production.