Journals
  Publication Years
  Keywords
Search within results Open Search
Please wait a minute...
For Selected: Toggle Thumbnails
A Multi-information Perception Based Method for Question Answering in Multi-party Conversation
GAO Xiaoqian, ZHOU Xiabing, ZHANG Min
Acta Scientiarum Naturalium Universitatis Pekinensis    2023, 59 (1): 21-29.   DOI: 10.13209/j.0479-8023.2022.069
Abstract464)   HTML    PDF(pc) (750KB)(121)       Save
Question answering in multi-party conversation typically focuses on exploring discourse structures or speaker-aware information but ignores the interaction between questions and conversations. To solve this problem, a new model which integrates various information is proposed. In detail, to hierarchically model the discourse structures, speaker-aware dependency of interlocutors and question-context information, the proposed model leverages above information to propagate contextual information, by exploiting graph convolutional neural network. Besides, the model employs a reasonable interaction layer based on attention mechanism to enhance the understanding of multi-party conversations by selecting more helpful information. Furthermore, the model is the first to pay attention to the explicit interaction between question and context. The experimental results show that the model outperforms multiple baselines, illustrating that the model can understand the conversations more comprehensively.
Related Articles | Metrics | Comments0
Data Augmentation Method for Question Answering
DING Jiajie, XIAO Kang, YE Heng, ZHOU Xiabing, ZHANG Min
Acta Scientiarum Naturalium Universitatis Pekinensis    2022, 58 (1): 54-60.   DOI: 10.13209/j.0479-8023.2021.112
Abstract648)   HTML    PDF(pc) (398KB)(162)       Save
Aiming at the problem that the current data augmentation method for automatic question answering requires a large amount of external data, a new method oriented to the defects of the question answering model is proposed. Firstly, the question answering (QA) model, question generating (QG) model and question answering matching (QAMatch) model are trained on the training set. Secondly, all the answers predicted by the QA model on the training set are obtained and the wrong ones are selected. Then, the QG model is used to generate corresponding questions for these answers. Finally, the question-answer pairs are filtered by the QAMatch model and the high-quality data are retained as the final augmented data. This method does not require additional data and domain knowledge, and can construct specific data for QA model, improving the performance with less training cost. Experimental results show that the proposed data augmentation method is effective for R-NET, Bert-Base and Luke. Compared with other methods, the QA model achieves better performance improvement with less data scale.
Related Articles | Metrics | Comments0
Distributions and Risk Assessment of Organochlorine Pesticides (OCPs) in Tanghe Groundwater
ZHANG Min, WANG Ting, YANG Chao, NI Jinren
Acta Scientiarum Naturalium Universitatis Pekinensis    2021, 57 (2): 283-290.   DOI: 10.13209/j.0479-8023.2021.014
Abstract613)   HTML    PDF(pc) (1475KB)(70)       Save
In order to evaluate the pollution status of organochlorine pesticides (OCPs) in the groundwater around Tanghe Wastewater Reservoir in Xiong’an New area, the authors analyzed OCPs residues in 36 groundwater samples collected in 2019. The composition characteristics and sources of HCHs and DDTs were preliminarily determined, and their health risks were assessed. The results suggested that Tanghe groundwater showed lower OCPs pollution with total amount of nd–45.60 ng/L and a detection rate of 77.78%, of which HCHs was the greatest distributor. The lower OCPs contents were partially attributed to implementations of the ecological sewage treatment project in 2018. Furthermore, both HCHs and DDTs were mainly historical residues originated from agricultural source, and DDTs probably decomposed into aerobic metabolite DDE. Overall, the noncarcinogenic and carcinogenic risks of OCPs were very low, posing no threat to human health though infants were more sensitive.
Related Articles | Metrics | Comments0
Neural Network Coupled Model for Conversion and Exploitation of Heterogeneous Lexical Annotations
HUANG Depeng, LI Zhenghua, GONG Chen, ZHANG Min
Acta Scientiarum Naturalium Universitatis Pekinensis    2020, 56 (1): 97-104.   DOI: 10.13209/j.0479-8023.2019.098
Abstract800)   HTML    PDF(pc) (673KB)(119)       Save
In order to expand the scale of manual annotated data and thereby improve model performance, we attempt to make full use of existing heterogeneous annotations to learn model parameters. We extend coupled sequence labeling model proposed by Li et al. (2015) under the BiLSTM-based deep learning framework. The neural coupled model learn its parameters directly on two heterogeneous training data, and predicts two optimal sequences simultaneously during the test phase. A lot of experiments have been conducted on the part-of-speech (POS) tagging task and the joint word segmentation and POS (WS&POS) tagging task. The results show that neural coupled approach is superior to other methods for exploiting heterogeneous lexical data, including the multi-task learning method and the traditional discrete-feature coupled model. Neural coupled model achieves higher performance on both scenarios, i.e., annotation conversion and boost the final target-side tagging accuracy by exploiting heterogeneous data.
Related Articles | Metrics | Comments0
Syntax-Enhanced UCCA Semantic Parsing
JIANG Wei, LI Zhenghua, ZHANG Min
Acta Scientiarum Naturalium Universitatis Pekinensis    2020, 56 (1): 89-96.   DOI: 10.13209/j.0479-8023.2019.099
Abstract1147)   HTML    PDF(pc) (644KB)(126)       Save
Considering the close correlation between syntactic and semantic structures, this paper attempts to add syntactic information into the universal conceptual cognitive annotation (UCCA) semantic parsing model to enhance the performance of semantic parsing. Based on the state-of-the-art graph-based UCCA semantic parser, we propose and compare four different approaches for incorporating syntactic information. Experiments are conducted on the English benchmark dataset for the semantic parsing shared task of the SemEval-2019 conference. The results on both the in-domain and out-domain evaluation data show that syntax-enhanced methods can achieve significant improvements of UCCA parsing. After utilizing BERT, syntactic information is still beneficial to some extent.
Related Articles | Metrics | Comments0
Characteristics of Humic Substances in KBD-Affected Region of Changdu, Tibet Based on PARAFAC of Fluorescence
JIANG Yong, GAO Dingxue, MAO Xuewen, YUAN Hao, HU Mingming, ZHANG Min, GUO Yongzhao, YI Malan, WU Jiang, XU Nan
Acta Scientiarum Naturalium Universitatis Pekinensis    2019, 55 (4): 717-726.   DOI: 10.13209/j.0479-8023.2019.031
Abstract1212)   HTML    PDF(pc) (4474KB)(124)       Save
Humic substances (HS) in Kaschin-Beck disease (KBD) affected regions were analyzed based on parallel factor analysis (PARAFAC) of fluorescence. Total organic carbon content of HS did not show significant difference between endemic and disease-free areas. Five fluorescence components were identified with PARAFAC, i.e. component 1 (oxidized quinone-like), component 2 (tryptophan-like), component 3 (terrestrial humic-like), component 4 (reduced quione-like) and component 5 (tyrosine-like). Component 1 (p<0.10), component 4 (p<0.05) of aquatic fulvic acid (FA) and component 4 (p<0.10) of aquatic humic acid (HA) in endemic areas showed higher content than disease-free areas. Lager differences of the quinone redox system in aquatic HS between endemic and disease-free areas exhibited in reduced quinone state than that in oxidized quinone state, and in FA than that in HA. HA showed higher content of reduced quinone than FA, but smaller differences between endemic and disease-free areas for its weaker influence on KBD due to extremely low carbon content in drinking water. Sediment HS showed mutual transformation with aquatic HS and higher content of reduced quinone, but no significant differences between endemic and disease-free areas. Intensive understanding on the differences of different fractions of HS and different state of quinone between endemic and disease-free areas can help guiding water improvement project in endemic areas.
Related Articles | Metrics | Comments0
Hypernym Relation Classification Based on Word Pattern
SUN Jiawei, LI Zhenghua, CHEN Wenliang, ZHANG Min
Acta Scientiarum Naturalium Universitatis Pekinensis    2019, 55 (1): 1-7.   DOI: 10.13209/j.0479-8023.2018.055
Abstract1347)   HTML    PDF(pc) (4709KB)(315)       Save

The authors propose a hypernym relation classification method based on word pattern, which can effectively alleviate the sparsity problem suffered by the traditional path-based method. Furthermore, this paper makes an effective combination of the path-based method and the distributional method via word pattern embedding. To demonstrate the effectiveness of the proposed approach, the authors manually annotated a Chinese hypernym dataset containing 12000 word pairs. The experimental results show that the proposed word pattern embedding approach is effective and can achieve an F1 score of 95.36%.

Related Articles | Metrics | Comments0
Translation Similarity Model Based on Bilingual Compositional Semantics
WANG Chaochao,XIONG Deyi,ZHANG Min
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract760)      PDF(pc) (511KB)(384)       Save
The authors propose a translation similarity model based on bilingual compositional semantics to integrate the bilingual semantic similarity feature into decoding process to improve translation quality. In the proposed model, monolingual compositional vectors for phrases are obtained at the source and target side respectively using distributional approach. These monolingual vectors are then projected onto the same semantic space and therefore transformed into bilingual compositional vectors. Base on this semantic space, translation similarity between source phrases and their corresponding target phrases is calculated. The similarities are integrated into the decoder as a new feature. Experiments on Chinese-to-English NIST06 and NIST08 test sets show that the proposed model significantly outperforms the baseline by 0.56 and 0.42 BLEU points respectively.
Related Articles | Metrics | Comments0
Research Advances on Xylose-Fermenting Recombinant Saccharomyces cerevisiae
ZHU Shunni,ZHANG Mingting,ZHOU Weizheng,ZHUANG Xinshu,XU Huijuan,QI Wei,YUAN Zhenhong
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract672)            Save
Bioconversion of lignocellulosic raw materials into ethanol has aroused extensive attention in recent years. High ethanol yields are dependent on co-fermentation of hexose and pentose sugars present in lignocellulosic hydrolysates. However the traditional ethanologenic yeast Saccharomyces cerevisiae has been found to be unable to ferment xylose and also unable to utilize this pentose sugar for growth. Hence attempts have been made to construct various recombinant yeasts to improve the ability of xylose fermentation. In this article, progress in development of xylose-fermenting recombinant S. cerevisiae strains is reviewed, including naturally occurring xylose-fermenting microorganisms, xylose metabolic pathway, metabolic engineering of S. cerevisiae, protoplast fusion and genome shuffling. Problems existing in current researches and prospects are also discussed.
Related Articles | Metrics | Comments0
Characteristics and Impact Factors of Litter Quantity and Development in Green Space of Shenzhen
ZHANG Xuan,TANG Qinglong,ZHANG Mingjie,ZENG Hui
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract607)            Save
The authors analyze the characteristics of the quantity and development of plant litter in Shenzhen, the correlation of standing stocks of different green-space type with the coverage of arboreal, shrub and herb, the edge disturbance, the frequency of treatment and the distance from the nearest road. Sanitation Department and Green Office are investigated to analyze the litter management and treatments. The results show that the highest quantity and best development of litter is found in the urban green-space type of natural forest, the productive green space and road greenbelt are in the next place; the worst is the roadside trees. Multiple correlation analysis show that the standing stocks of existing litter is highly related with the frequency of treatment and the coverage of shrub. Multiple Linear Regression Models including all influencing factors of different green-space types can explain 40.6%-93.3% of the variation in standing stocks. Green spaces in Shenzhen are arranged by the human-impact level on their litter from the worst to best as: roadside trees; the village green space, unit attached green space, park green space and community green space; the productive forest and road greenbelt; the natural forest.
Related Articles | Metrics | Comments0
Soil Wind Erosion under Different Land Use Types in Bashang of Hebei Province
SUN Yanrong,LIU Hongyan,FAN Tao,MA Liguo,ZHANG Mingjie
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract540)            Save
By the trace of 137Cs and soil granularity analysis, the relationship between land-use and soil wind erosion was investigated in Bashang of Hebei Province. The results indicate that the land use types with erosion intensity, from the higher to the lower, are sandy hilly grassland > cultivated land > grassland converted from cultivated land > artificial forest > gravel hill grassland. Hence, to recover the vegetation, the gravel hilly grassland in this region should be reserved and the original vegetation in it should be protected.
Related Articles | Metrics | Comments0
Research on the Change of Motorized Passenger Transport Energy Consumption in Selected Chinese Cities Based on LMDI Method
ZHANG Ming,MU Hailin
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract711)            Save
The change of urban passenger transport energy consumption is influenced by many factors, such as urbanization rate, resident travel behaviors etc. To better understand the dominant factors that influence the energy consumption, it is important to analyze the driving force quantitatively. Here, a new decomposition model based on LMDI method is developed to analyze China's passenger transport energy in mega-cities. Empirical analysis of the model is applied to Beijing, Shanghai and Guangzhou. The results show that most factors lead to the growth of urban passenger transport energy consumption in various degrees.
Related Articles | Metrics | Comments0
Innovation of Administrate System of Science-Technological Research Results in Universities
WANG Jin,ZHANG Ping,ZHOU Hui,ZHANG Ming
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract495)            Save
The authors discuss how to establish a system to administrate the science-technology research results and make it useful and helpful to facilitate the application of science-technology research results, which is in harmony with market economy. The authors also discuss how people can put into practice the pattern of administrating the science-technology research results adopted by U.S. universities and how to administrate the technology transfer in the market economy. In addition, the authors discuss the technology transfer from the perspectives of the evaluation, registration and secrecy of the science-technology research results, the application of the patent, and the protection of intelligence property.
Related Articles | Metrics | Comments0
Design and Implementation of MCGE Issue's Calculation Based on Webservice
YU Hailong,WU Lun,LIN Xin,QIN Shi,Zhang Ming
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract653)            Save
At present, the calculation of the Modern Catchment Geomorphic Evolution (MCGE) issues is mainly solved by systems' integration, which results in function redundancy, lower efficiency and more difficult modification. The method to resolve the MCGE issues based on WebService is proposed. The design of the Modern Catchment Geomorphic Evolution Model Service(MCGEMS), and integration between MCGEMS and WCS are given. Within .NET interface, the MCGEMS and integration with WCS are implemented respectively. This method resolves the problems resulting from systems integration, improves the MCGEM reuse capability and simplifies the procedure of integration.
Related Articles | Metrics | Comments0
Overview of Ontology
DENG Zhihong,TANG Shiwei,ZHANG Ming,YANG Dongqing,CHEN Jie
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract963)            Save
Ontology is defined as an explicit formal specification of a shared conceptualization. It can provide semantic meaning through relations between concepts. As a fine model for presenting hierarchy and semantic meaning of concepts, Ontology is widely concerned and extensively applied to many fields in computer science and technology. With regard to little research on ontology in China, The state of the art of ontology is surveyed in this paper. This work first analyzes connotation and methodology of ontology, and then analyzes its applications in information system in details. The paper ends with a short conclusion and future work.
Related Articles | Metrics | Comments0
Trends and Spatial Distributions of Methane Emissions in Beijing
SU Fang,SHAO Min,ZHANG Min
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract682)            Save
Based on the methodology guidelines of IPCC, a city-scale methane emission inventory for Beijing in 1999 was established. The statistical data were derived from newly intensive survey and updated emission factors were referenced. Results showed that the total methane emissions were 296.29 Gg. The main sources were municipal solid waste and fossil fuel, which produced 161.60 Gg and 96.08 Gg of methane emissions, accounting for 54.6% and 32.4% of total emissions respectively. The spatial distributions of methane emission source strengths were also obtained.
Related Articles | Metrics | Comments0
Study on the Effective Composition of Skimmia laureola
ZHANG Hongjie,ZHANG Mingzhe
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract628)            Save
Five compounds were separated from the herb plant Skimmia laureola ssp. multinervia collected from Yunnan Province of China. They were identified as O-methyllaureolol(I), taraxerone(II), (-)-evoxine(III), kokusaginine(IV) and haplopine(V) by UV, IR, 1H-NMR and 13C-NMR, in which compound I is a new compound.
Related Articles | Metrics | Comments0
Studies on the Chemical Constituents of the Parasite of Rhododendron Delavayi Franch
ZHANG Qihan,GUO Jian,ZHANG Mingzhe
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract713)            Save
The parasite of Rhododendron delavayi Franch, collected from Yunnan Province, was extracted by organic solvents with different polarity. After chromatography and purification six crystallines were isolated from this plant for the first time. On the bases of physical and chemical properties and spectral analyses five of them were identified as nonacosane(A), succinic acid(C), quercetin(D), quercetin-3-O-α-arabinofuranoside(E), and quercetin-3-O-α-rhamnopyranoside(F) separately, crystalline B, identified by GC/MS, is a mixture composed of campesterol(Ⅰ), brassicasterol(Ⅱ), stigemasterol(Ⅲ) and γ -sitosterol(Ⅳ).
Related Articles | Metrics | Comments0
Studies on the Constituents of Flavonoids from Craibiodendron Yunnanense
XU Baoan,SHU Ye,ZHANG Mingzhe
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract657)            Save
The leaves of Craibiodendron yunnanense W.W. Smith collected from Yunnan Province were extracted by organic solvents with different polarity. After chromatography for many times and purification, six compounds were isolated for the first time from this plant. On the basis of determination of physical and chemical properties and spectral analyses they were identified as 11 α-hydroxy-α-amyrin(Ⅰ), β-sitosterol(Ⅱ), quercetin(Ⅲ),quercetin-3-O-α-arabinofuranoside(Ⅳ), quercetin-3-O-α-rhamnopyranoside(Ⅴ), quercetin-3-O-β-galactopyranoside(Ⅵ). The pharmacological verification of crude flavonoids was made preliminarily on mice, the results showed it has central inhibition.
Related Articles | Metrics | Comments0
A Quantitative Discussion to the Relationship between Thrusting of Orogenic Belt and Subsidence and Sedimentation of Foreland Basin ——Kuche Intracontinental Foreland Basin as an Example
ZHANG Mingshan,QIAN Xianglin,LI Maosong
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract823)            Save
The basin between orogenic belt and craton is called foreland basin. However, in a sense it has been expanded in the last few years because of development of the intracontinental orogenic theory. The authors suggest that the basin which is located at the front of the intracontinental orogenic belt may also be called foreland basin or intracontinental foreland basin because it can be analogic with the typical foreland basin. The thrusting of the orogenic belt onto the foreland area is an indicator of the tectonic response to lithosphere in foothill, and therefore, the subsidence(foreland basin) and the nearby uplift(Mountain) are formed. Besides common features, each foreland basin has some local features differing from the influence of the basement character, tectonic setting, thrusting of orogenic belt, growing rate of denudation and sedimentation, and open extent of the basin. The newestapproach of the study is the balanced relationship in foreland basin reflectedby the tectonic regime and sedimentation between thrusting of the orogenic belt and the foreland basin, shown by the curve of the corrected tectonic stripping, time step and rate of sedimentation. The Kuche intracontinental foreland basin, located at southern foothill of south Tianshan is a typical closed system for this type of tectonsedimentary tary processes The studyno the Kuche intracontinental foreland basin using the methods mentioned above shows that corrected curves of the tectonic stripping can describe the thrusting and subsidence accurately. Those results of the study could be valuable for evaluation of the oilbearing reservoir and structural trap based on the regional geological evidences in the Kuche intracontinental foreland basin.
Related Articles | Metrics | Comments0
Studies on Volatile Constituents from Skimmia Laureola ssp. Multinervia
ZHANG Hongjie,GUAN Ningning,ZHANG Mingzhe
Acta Scientiarum Naturalium Universitatis Pekinensis   
Abstract721)            Save
Volatile constituents of components A and B aquired from Skimmia laureola ssp. multinervia by extraction with petroleum ether was investigated by GC/MS method. GC/IR technique was applied for A as a complementary means. Eighteen compounds were identified such as neophytadiene, methyl laurate etc.
Related Articles | Metrics | Comments0