[an error occurred while processing this directive]

Table of Content

    20 January 2021, Volume 57 Issue 1
    Abstractive Text Summarization Based on Semantic Alignment Network
    WU Shixin, HUANG Degen, LI Jiuyi
    2021, 57(1):  1-6.  DOI: 10.13209/j.0479-8023.2020.084
    Asbtract ( )   HTML   PDF (503KB) ( )  
    Related Articles | Metrics
    Aiming at the problem of insufficient utilization of the overall semantic information of abstracts in decoding by the currently abstractive summarization model, this paper proposes a neural network automatic abstract model based on semantic alignment. This model is based on the Sequence-to-Sequence model with attention, Pointer mechanism and Coverage mechanism. A semantic alignment network is added between the encoder and the decoder to achieve the semantic information alignment of the text to the abstract. The achieved semantic information is concatenated with the context vector in decoding, so that when the decoder predicts the vocabulary, it not only uses the partial semantics before decoding, but also considers the overall semantics of the digest sequence. Experiments on the Chinese news corpus LCSTS show that the proposed model can effectively improve the quality of abstractive summarization.
    A Multi-modal Sentiment Recognition Method Based on Multi-task Learning
    LIN Zijie, LONG Yunfei, DU Jiachen, XU Ruifeng
    2021, 57(1):  7-15.  DOI: 10.13209/j.0479-8023.2020.085
    Asbtract ( )   HTML   PDF (3483KB) ( )  
    Related Articles | Metrics
    In order to learn more emotionally inclined video and speech representations through auxiliary tasks, and improve the effect of multi-modal fusion, this paper proposes a multi-modal sentiment recognition method based on multi-task learning. A multimodal sharing layer is used to learn the sentiment information of the visual and acoustic modes. The experiment on MOSI and MOSEI data sets shows that adding two auxiliary single-modal sentiment recognition tasks can learn more effective single-modal sentiment representations, and improve the accuracy of sentiment recognition by 0.8% and 2.5% respectively.
    Robustness of Chinese Machine Reading Comprehension
    LI Yeqiu, TANG Hongxuan, QIAN Jin, ZOU Bowei, HONG Yu
    2021, 57(1):  16-22.  DOI: 10.13209/j.0479-8023.2020.088
    Asbtract ( )   HTML   PDF (454KB) ( )  
    Related Articles | Metrics
    In order to better evaluate the robustness of Machine Reading Comprehension (MRC) models, this paper builds three test sets from Dureader by automatically extracting and manually annotating, consisting of oversensitivity, over-stability, and generalization. In addition, this paper proposes a multi-task learning framework with answer extraction task and masked position prediction task. Experimental results demonstrate that proposed method gains significant robustness improvements and show the effectiveness of the three test sets on evaluating the robustness of MRC models.
    Abstractive Summarization Based on Fine-Grained Interpretable Matrix
    WANG Haonan, GAO Yang, FENG Junlan, HU Min, WANG Huixin, BAI Yu
    2021, 57(1):  23-30.  DOI: 10.13209/j.0479-8023.2020.082
    Asbtract ( )   HTML   PDF (850KB) ( )  
    Related Articles | Metrics
    According to the great challenge of summarizing and interpreting the information of a long article in the summary model. A summary model (Fine-Grained Interpretable Matrix, FGIM), which is retracted and then generated, is proposed to improve the interpretability of the long text on the significance, update and relevance, and then guide to automatically generate a summary. The model uses a pair-wise extractor to compress the content of the article, capture the sentence with a high degree of centrality, and uses the compressed text to combine with the generator to achieve the process of generating the summary. At the same time, the interpretable mask matrix can be used to control the direction of digest generation at the generation end. The encoder uses two methods based on Transformer and BERT respectively. This method is better than the best baseline model on the benchmark text summary data set (CNN/DailyMail and NYT50). The experiment further builds two test data sets to verify the update and relevance of the abstract, and the proposed model achieves corresponding improvements in the controllable generation of the data set.
    Multi-Turn Conversation Rewriter Model Based on Masked-Pointer
    YANG Shuangtao, FU Bo, YU Chenchen, HU Changjian
    2021, 57(1):  31-37.  DOI: 10.13209/j.0479-8023.2020.087
    Asbtract ( )   HTML   PDF (764KB) ( )  
    Related Articles | Metrics
    To solve the problem of Non-Sentential Utterances in multi-turn conversations, Masked Rewriter Model is proposed based on the Masked Language Model, and the rewriting performance is significantly improved compared with the Seq2Seq-based rewriting model. Considering the NSUs rewriting task characteristics, Masked-Pointer Rewriter Model is proposed based on the Masked Language Model and Pointer Network, which achieves better rewriting results than the Masked Rewriter Model by using the Pointer Network to enhance the model’s attention to historical information.
    An Antinoise Response Generation for Open Domain Dialogue System
    ZHU Qinpei, MIAO Qingliang
    2021, 57(1):  38-44.  DOI: 10.13209/j.0479-8023.2020.089
    Asbtract ( )   HTML   PDF (694KB) ( )  
    Related Articles | Metrics
    In order to reduce the noise interference on the response generation model, this paper proposes an antinoise model based on encoder-decoder architecture. Firstly, simulation noisy characters are added to the input utterances. Then noisy character recognition is trained at the encoder output layer, thus improving the ability of extracting noise features. Finally, pre-trained language model is fused at the encoder output layer to expand the coverage of noise. An antinoise test set is presented for verifying the model’s antinoise effect, which is the first Chinese single-turn open domain dialog system corpus with real noise. Experiments show that the proposed model’s results of automatic evaluation and manual evaluation on the antinoise test set are better than the baseline models.
    Syntax-based Code Generation Model with Selective Local Attention and Pre-order Information LSTM Decoder
    LIANG Wanying, ZHU Jia, WU Zhijie, YAN Zhiwen, TANG Yong, HUANG Jin, YU Weihao
    2021, 57(1):  45-52.  DOI: 10.13209/j.0479-8023.2020.086
    Asbtract ( )   HTML   PDF (805KB) ( )  
    Related Articles | Metrics
    This paper proposes a syntax-based code generation model with selective local attention and a preorder information decoder based on long-short term memory (LSTM) neural network, which aims to enhance the relevance by changing the calculation scope of the context vector and fuse more pre-order information during the decoding process. Code generation experiments in two dataset Hearthstone and Django confirm the effectiveness of this model. Compared with state-of-the-art models, the proposed model not only achieves excellent accuracy and bilingual evaluation understudy score, but also minimizing computational effort.
    Joint Extraction of Entities and Relations Based on Hierarchical Sequence Labeling
    TIAN Jialai, LÜ Xueqiang, YOU Xindong, XIAO Gang, HAN Junmei
    2021, 57(1):  53-60.  DOI: 10.13209/j.0479-8023.2020.083
    Asbtract ( )   HTML   PDF (1050KB) ( )  
    Related Articles | Metrics
    In order to further improve the effect of entity relationship joint extraction, this paper proposes an endto-end joint extraction model (HSL). HSL model adopts a new labeling scheme to transform the joint extraction of entities and relationships into sequence labeling problems, and uses a layered sequence labeling method to solve the problem of triple overlap. The experiments demonstrates that HSL model can effectively deal with the problem of triple overlap and improve the extraction effect. The F1 value on the military corpus data set reaches 80.84%, and 86.4% on the WebNLG open data set, which exceeds the current mainstream triple extraction model, improving the effect of triple extraction.
    Chinese Spelling Correction Method Based on Transformer Local Information and Syntax Enhancement Architecture
    DUAN Jianyong, YUAN Yang, WANG Hao
    2021, 57(1):  61-67.  DOI: 10.13209/j.0479-8023.2020.081
    Asbtract ( )   HTML   PDF (874KB) ( )  
    Related Articles | Metrics
    Two new methods for improving Chinese spelling correction are proposed. The first one is to add Gaussian Bias matrices to the Transformer’s attention mechanism, which is used to improve the model’s attention to local text and to extract information from the wrong words and the surrounding text in the error text. Secondly, the ON_LSTM model is used to extract grammatical information on the special grammatical structure features exhibited by the error text. The experimental results show that both methods are effective in improving accuracy and recall, and the model after fusing the two methods achieves the highest F1 value.
    Research on the Construction and Application of Paraphrase Parallel Corpus
    WANG Yasong, LIU Mingtong, ZHANG Yujie, XU Jin’an, CHEN Yufeng
    2021, 57(1):  68-74.  DOI: 10.13209/j.0479-8023.2020.078
    Asbtract ( )   HTML   PDF (398KB) ( )  
    Related Articles | Metrics
    Taking Chinese as the research object, the authors put forward the method to construct large-scale and high-quality paraphrase parallel corpora. The paraphrase data augmentation method include transfering English paraphrase corpus to Chinese, by using the method of translation engines, and manually annotating evaluation data set. Based on the constructed Chinese paraphrase data, the validity of the paraphrase data construction application method is verified in the paraphrase recognition task and natural language inference task. Firstly, the paraphrase recognition data is generated based on the constructed paraphrase corpus, and the attention-based neural network model of sentence matching is pre-trained to capture the paraphrase information. Then, the pre-trained model is applied to the natural language inference task to improve the performance. The experimental results on the open set show that the constructed paraphrase corpus can be effectively applied to the paraphrase recognition task, and the model can learn paraphrase knowledge. When applied to natural language inference task, paraphrase knowledge can effectively improve the accuracy of natural language inference models and verify the effectiveness of paraphrase knowledge for downstream semantic understanding tasks. Meanwhile, the proposed construction method for the paraphrase corpus is language-independent, which can provide more training data for other languages and fields, generate high-quality paraphrase data, and further improve the performance of other tasks.
    Object Space Relation Mechanism Fused Image Caption Method
    WAN Zhang, ZHANG Yujie, LIU Mingtong, XU Jin’an, CHEN Yufeng
    2021, 57(1):  75-82.  DOI: 10.13209/j.0479-8023.2020.080
    Asbtract ( )   HTML   PDF (5984KB) ( )  
    Related Articles | Metrics
    Focusing on the specific information of the positional relationship between objects in the image, a neural network image summary generation model integrating spatial relationship mechanism is proposed, in order to provide key information (object position or trajectory) for downstream tasks such as visual question answering and voice navigation. In order to enhance the learning ability of the positional relationship between objects of the image encoder, the geometric attention mechanism is introduced by improving the Transformer structure, and the positional relationship between objects is explicitly integrated into the appearance information of the objects. In order to assist in the completion of specific information-oriented extraction and summary generation tasks, a data production method for relative position relations is further proposed, and the image abstract data set Re-Position of the position relations between objects is produced based on the SpatialSense data set. The experimental results of comparative evaluation with five typical models show that the five indicators of the proposed model are better than those of other models on the public test set COCO, and all six indicators are better than those of other models on Re-Position data set.
    Unsupervised Syntactically Controllable Paraphrase Network for Adversarial Example Generation
    YANG Erguang, LIU Mingtong, ZHANG Yujie, MENG Yao, HU Changjian, XU Jin’an, CHEN Yufeng
    2021, 57(1):  83-90.  DOI: 10.13209/j.0479-8023.2020.079
    Asbtract ( )   HTML   PDF (570KB) ( )  
    Related Articles | Metrics
    Prior work on adversarial example generation with syntactically controlled paraphrase networks requires large-scale paraphrase parallel corpora to train models. The performance of the model is seriously limited by the domain and scale of paraphrase parallel corpus. To solve this problem, this paper proposes an unsuprervised syntactically controlled paraphrase model to generate adversarial examples which only needs monolingual data. Specifically, variational autoencoder is used to learn model, which maps a sentence and a syntactic parse tree into semantic and syntactic variables, respectively. By learning to reconstruct the input sentence from syntactic and semantic variables, the model effectively learns to generate syntactic paraphrases without using any parallel data. Experiment results on unsupervised sentence paraphrasing and adversarial example generation demonstrate that the proposed model achieves new state-of-the-art results on unsupervised paraphrase generation and generate effective adversarial examples. These examples can be used to improve the robustness and generalization of NLP (natural language processing) model.
    A Review of Entity Linking Research Based on Deep Learning
    LI Tianran, LIU Mingtong, ZHANG Yujie, XU Jin’an, CHEN Yufeng
    2021, 57(1):  91-98.  DOI: 10.13209/j.0479-8023.2020.077
    Asbtract ( )   HTML   PDF (457KB) ( )  
    Related Articles | Metrics
    The authors introduce the concept and steps of entity linking in detail, and the problems and research status of named entity linking based on deep learning in recent years, analyze the problems and corresponding solution models of entity linking and present related data sets and evaluation methods. The authors summarize the current status of entity linking in international evaluation conferences and analyze the future research directions.
    Modeling for Dynamic Monitoring of Marine Gas Hydrate Exploitation Using 4C-OBC Time-lapse Seismic System
    ZHU He, HE Tao, LIANG Qianyong, WU Xuemin, DONG Yifei
    2021, 57(1):  99-110.  DOI: 10.13209/j.0479-8023.2020.091
    Asbtract ( )   HTML   PDF (17886KB) ( )  
    Related Articles | Metrics
    In order to control the geological and environmental risks during the exploitation of marine gas hydrate, 4-component ocean bottom cable (4C-OBC) is considered to perform time-lapse seismic monitoring on hydrate reservoirs, which can collect P- and S-wave simultaneously, and satisfy the requirement of real-time and long-term monitoring. This paper uses ray tracing method to carry out forward simulation of 4C-OBC time-lapse seismic system for the horizontal well environment in the future commercial gas hydrate exploitation. Based on the seismic illumination of the formation model, the optimal OBC layout parameters is obtained to ensure that the acquired seismic data has good imaging effect. Then, the travel time and amplitude of the time-lapse seismic data in different exploitation stages is analyzed. The results show that both differential travel time and amplitude could reflect the exploitation degree of gas hydrate reservoir, especially significant for converted S-wave. The error analysis results of the observation system show that seismic source vessel’s positioning error would not significantly affect the time-lapse monitoring system. In sum, it is effective to monitor dynamic process of marine gas hydrate reservoir using 4C-OBC time-lapse seismic system.
    Mineralogical Mechanism of Micro-Remaining Oil Occurrence: An Example Study of Middle-Low Permeability Sandstone Reservoir of Ordos Basin
    WANG Zhelin, SHI Yongmin, PAN Mao, WANG He, MA Zilin
    2021, 57(1):  111-120.  DOI: 10.13209/j.0479-8023.2020.116
    Asbtract ( )   HTML   PDF (35025KB) ( )  
    Related Articles | Metrics
    In order to figure out the existing problems of low visualization accuracy in the research of microresidual oil and clarify its morphological characteristics, this study discussed the distribution of residual oil occurrence state and its relationship with minerals. We conducted an integrated approach of core observation, casting film, X-ray diffraction (X-RD), field emission environmental scanning electron microscope (FE-SEM) and energy-dispersive spectrometry (EDS) analysis on middle-low permeability sandstone reservoir in the 9th member of Yanan Formation and 2nd member of Yanchang Formation from Ordos Basin, China. The result shows that different pore structure forms have different storage capacity for remaining oil, it depends on the properties of matrix minerals corresponding to the pore, including mineral morphology, surface roughness, specific gravity, wettability, etc, which can be classified into five types: residual oil block mess, semi-free oil blob, semi-free oil island, semi-free oil mist and irreducible oil.
    Coupled Model Studies of the Tibetan Plateau Effect on the Atlantic Meridional Overturning Circulation under Different Resolutions
    SHAO Xing, YANG Haijun, LI Yang, JIANG Rui, YAO Jie, YANG Qianzi
    2021, 57(1):  121-131.  DOI: 10.13209/j.0479-8023.2020.092
    Asbtract ( )   HTML   PDF (1934KB) ( )  
    Related Articles | Metrics
    The effect of Tibetan Plateau on the Atlantic Meridional Overturning Circulation (AMOC) under different resolutions is studied using the coupled Community Earth System Model (CESM1.0). Comparation of the results with and without the Tibetan Plateau tests shows that the changes of AMOC after the removal of the Tibetan Plateau are related to the resolution of the model. Under different resolutions, the changes of AMOC are inconsistent: AMOC Index decreases by 89% in the low resolution test, but only by 13% in the high resolution test. The reason for this difference is that there are significant differences in the changes of location and strength of the mixed layer subduction, which contributes to the deep water formation under different resolution test: the low resolution test is mainly located in the GIN seas, while the high resolution test is mainly located in the Labrador Sea. After removing the Tibetan Plateau, the subduction of both tests decreases, but the decrease of the low resolution test is larger than that of high resolution test. The subduction in the Labrador Sea of high resolution test decreases the most obviously, while the subduction in all sea areas decreases in the low resolution test, especially in the GIN seas. Comparation of the observed wind data and latest observational studies of deep water formation area over the North Atlantic shows that the results of low resolution coupled model are more similar to the actual observations in the seas studied in this paper.
    Impact of Overflow Pollution on Water Quality in Shenzhen Bay
    CHENG Peng, LI Mingyuan, LOU Kai, QIN Huapeng
    2021, 57(1):  132-142.  DOI: 10.13209/j.0479-8023.2020.097
    Asbtract ( )   HTML   PDF (10360KB) ( )  
    Related Articles | Metrics
    This study takes the Shenzhen Bay Basin as the research object, based on the combined model of rainfall runoff pollution, overflow from sewer interception system and hydrodynamic water quality in the bay, the temporal and spatial variation of overflow pollution and its effect on water quality in the bay were analyzed. The results show that in the whole year, the main pollution sources of the basin are the pollution of the wastewater treatment plant and the overflow pollution. The overflow pollution accounts for about 30% of the total pollution load, rises to about 50% in the rainy season, and becames the main pollution source in the rainy season in Shenzhen Bay Basin. The overflow pollution load in rainy season accounts for more than 85% of the total overflow pollution load in the whole year. The pollution load of overflow increases with the rainfall. When the rainfall intensity is similar, the longer the dry time before rain, the greater the overflow pollution load. In terms of water quality in the bay, the water quality in the Inner Bay is worse than that in the Outer Bay and the fluctuation is more obvious, In rainy season, the water quality in Inner Bay and Middle Bay is worse and fluctuates more than that in dry season. In the event of rain, the water quality fluctuation in the inner bay was most seriously affected by the impact of overflow pollution, while the middle and outer bays was mainly affected by the tidal action. When the return period of rainfall is 0.25 to 0.5 years, the water quality in Shenzhen Bay exceeds the standard most obviously, and the duration of water quality affected by overflow pollution is 12 to 20 days.
    Biogeographic Patterns of Microbial Communities Associated with Syntrophic Butyrate Degradation in Paddy Soils in Eastern China
    FEI Yuanyuan, JIAO Shuo, LU Yahai
    2021, 57(1):  143-152.  DOI: 10.13209/j.0479-8023.2020.109
    Asbtract ( )   HTML   PDF (1841KB) ( )  
    Related Articles | Metrics
    The authors collected 34 paddy soil samples along the latitude from eastern China. Enrichment experiment was conducted under anaerobic conditions with sodium butyrate as the sole substrate. The authors investigated the microbial community characteristics and the functional activity of syntrophic butyrate degradation and the biogeographic patterns of relative abundance of Syntrophomonas in these soils by Illumina sequencing of 16S rRNA genes. The lag phase of CH4 production (3–14 days) increased towards higher latitudes, whereas the maximum rate of CH4 production did not. The correlation analysis on influencing factors revealed that Syntrophomonas was the key syntrophic bacterial taxon associated with butyrate degradation and its relative abundance was significantly influenced by mean annual temperature (MAT). The sampling sites with a relatively high abundance of Syntrophomonas had a shorter time for the complete degradation of butyrate. Distance-decay patterns characterized by a steeper slope were found in microbial communities associated with syntrophic butyrate degradation (p<0.05), indicating that the construction of microbial communities was driven by both spatial distance and environmental factors.
    Change of NDVI during Growing Season and Its Relationship with Climate in North China and the Adjacent Areas from 1982 to 2014
    ZHANG Xinyue, FENG Yuhao, ZENG Hui, TANG Zhiyao
    2021, 57(1):  153-161.  DOI: 10.13209/j.0479-8023.2020.108
    Asbtract ( )   HTML   PDF (8736KB) ( )  
    Related Articles | Metrics
    Using data from 690 meteorological observatories and GIMMS NDVI 3g data from 1982 to 2014, trend analysis, wavelet partial cross-correlation analysis, partial correlation analysis and lag analysis were used to explore the change rule of NDVI during the growing season (May to October) and its relationship with climate in North China and the adjacent areas in the past 33 years. The results showed that the average growing season NDVI increased from 0.44 in the 1980s to 0.49 in the 2010s. NDVI in the growing season increased rapidly in the central part of the research area, but decreased in the northwest desert area. The increase of NDVI in the growing season of the research area was benefited from the increase of temperature and precipitation, and the influence of precipitation was greater. NDVI of the research area was positively correlated with the temperature in most areas. Except for the southeastern part of the study area, NDVI and precipitation had strong positive correlation. At 15-day resolution, the response of NDVI to temperature in the growing season in most areas did not have obvious lag or was lagged in one period (15 days), and the response to precipitation was lagged about 1–2 periods (15–30 days). Therefore, in general, vegetation growth in North China and the adjacent areas responded more rapidly to temperature than precipitation.
    Impact of Model Structure and Parameterization Differences on Evapotranspiration Estimation
    ZHAO Wenli, XIONG Yujiu, QIU Guoyu, YAN Chunhua, ZOU Zhendong, QIN Longjun
    2021, 57(1):  162-172.  DOI: 10.13209/j.0479-8023.2020.119
    Asbtract ( )   HTML   PDF (4836KB) ( )  
    Related Articles | Metrics
    Based on the HiWATER high-density eddy covariance (EC) tower observations in Heihe Oasis in 2012, the impact of model structure differences (comparison between one-source Penman-Monteith / PM equation and two-source PM equation, or comparison between two-source PM equation and two-source three-temperature model) and parameterization differences on the evapotranspiration estimation were evaluated. The results show that, 1) compared with the two-source PM equation with a relatively complex model structure, the mean absolute percent error (MAPE) estimated by the one-source PM equation is 34%, which is more accurate than that by the two-source PM equation (40%); 2) for two kinds of two-source model with significant differences in model structure, the three-temperature model without resistance parameters has higher estimation accuracy than the PM-based equation with resistance parameters. The former has a MAPE of 18% (R2=0.85), while the PM-based equation has that of 40% (R2=0.34); 3) two one-source and one two-source resistance parameterization methods lead to different evapotranspiration estimation accuracy for the PM-based equation, with a MAPE difference of up to 6%; 4) using prior knowledge / dataset to calibrate resistance parameterization can significantly improve the estimation accuracy of one-source PM equation (MAPE can be reduced by 22%), but as model structure and parameterization complexity increase, two-source PM equation hasn’t been improved significantly after resistance parameterization calibration (MAPE is only reduced by 0.8%).
    Evaluation and Analysis of Ecosystem Services Value in Beijing-Tianjin-Hebei Region Based on Demand Zoning
    TANG Xiumei, LIU Yu, REN Yanmin, ZHOU Yanbing
    2021, 57(1):  173-180.  DOI: 10.13209/j.0479-8023.2020.112
    Asbtract ( )   HTML   PDF (945KB) ( )  
    Related Articles | Metrics
    Based on the analysis of the logical relationship between supply and demand of ecosystem services, this study evaluates the demand status of ecosystem services in Beijing-Tianjin-Hebei region, delimits the demand type areas, calculates the spatial and temporal changes of ecosystem services value (ESV) based on land use status maps in 2000 and 2015, and puts forward corresponding land use strategies. The result is as follows. 1) Human beings have material, environmental and cultural needs for ecosystem. The demand for ecosystem services can be evaluated from four aspects: population, economic level, industrial development and educational level, corresponding to the nine services of the four functions of ecosystem, including supply, regulation, support and culture; 2) There is a large gap in the demand for ecosystem services in Beijing-Tianjin-Hebei region, which can be divided into four types: extremely high demand area, high demand area, medium demand area and low demand area. 3) From 2000 to 2015, the total value of ecosystem services in Beijing-Tianjin-Hebei region decreased. At county level, the total value of ecosystem services and the average value of land decreased gradually from north to south in space; 4) The value distribution of ecosystem services in different demand areas was unbalanced. From 2000 to 2015, the value of all types of areas has decreased, and the land use strategies of different types of areas are different.
    Research on China’s Carbon Dioxide Emissions Efficiency from 2007 to 2016: Based on Two Stage Super Efficiency SBM Model and Tobit Model
    NING Lunchen, ZHENG Wen, ZENG Liang’en
    2021, 57(1):  181-188.  DOI: 10.13209/j.0479-8023.2020.111
    Asbtract ( )   HTML   PDF (811KB) ( )  
    Related Articles | Metrics
    The efficiency of carbon emissions of 30 provincial regions (Tibet Autonomous Region and Hong Kong, Macao and Taiwan are not included) in China from 2007 to 2016 was studied by using SE-SBM model, and the factors that influence the efficiency were analyzed by using Tobit model. The conclusions are as follows. Since 2007, China’s efficiency of carbon emissions has slightly decreased, and slightly improved after 2015, while the regional difference is obvious, and Beijing is the most efficient region. At the four region, eastern area has the high-test level, and it has significant difference compared with western area which has the lowest level in carbon emission efficiency level. However, the efficiency of carbon emissions of western area has been gradually increasing in recent years, and it has the trend of matching the central and northeastern regions’ efficiency of carbon emissions. By Tobit regression analysis, the results show that government interventions, energy intensity, level of opening-up, energy structure and technical level have an impact on the efficiency of carbon emissions. At last, the paper suggests that government should allocate carbon emissions targets with considering the differences in regional industrial structure and the level of economic development, improve opening-up structure and upgrade energy structure.
    Impact of Environmental Regulation on Industy Green Total Factor Productivity: Mediating Effect of Short-term Liquidity
    LIU Jinhui, ZOU Zhendong, QIU Guoyu
    2021, 57(1):  189-198.  DOI: 10.13209/j.0479-8023.2020.122
    Asbtract ( )   HTML   PDF (752KB) ( )  
    Related Articles | Metrics
    Based on system-generalized method of moments (SYS-GMM), this paper analyzes the influence of environmental regulation on the GTFP (green total factor productivity) of 37 industries in China during the period of 2003–2015 by mearsuring the intermediate effect of liquidity ratio. The result shows that the compound annual growth rate (CAGR) of GTFP among all industries was 6.8%. Technical development was the main contributor ameliorating GTFP. GTFP of industries above designated size goes upward and then downward in accordance with the stringency of environmental regulation. Short-term liquidity plays a role as intermediate variable, since it drops as environmental regulation gets sticter, which leads to a rise in GTFP of industries. When environmental regulation gets too strict for enterprises to comply with, their short-term liquidity ratio decreases significantly, which brings financial risks and leads to a reduction of GTFP. These conclusions are of great significance for government and enterprises to enact and comply with environmental and development strategies.