Text Simplification Using Transformer and BERT

Reading and writing are the main interaction methods with web content.Text simplification tools are helpful for people with cognitive impairments,new language learners,and children as they might find difficulties in understanding the complex web content.Text simplification is the process of changing complex text intomore readable and understandable text.The recent approaches to text simplification adopted the machine translation concept to learn simplification rules from a parallel corpus of complex and simple sentences.In this paper,we propose two models based on the transformer which is an encoder-decoder structure that achieves state-of-the-art(SOTA)results in machine translation.The training process for our model includes three steps:preprocessing the data using a subword tokenizer,training the model and optimizing it using the Adam optimizer,then using the model to decode the output.The first model uses the transformer only and the second model uses and integrates the Bidirectional Encoder Representations from Transformer(BERT)as encoder to enhance the training time and results.The performance of the proposed model using the transformerwas evaluated using the Bilingual Evaluation Understudy score(BLEU)and recorded(53.78)on the WikiSmall dataset.On the other hand,the experiment on the second model which is integrated with BERT shows that the validation loss decreased very fast compared with the model without the BERT.However,the BLEU score was small(44.54),which could be due to the size of the dataset so the model was overfitting and unable to generalize well.Therefore,in the future,the second model could involve experimenting with a larger dataset such as the WikiLarge.In addition,more analysis has been done on the model’s results and the used dataset using different evaluation metrics to understand their performance.

An Efficient Long Short-Term Memory Model for Digital Cross-Language Summarization

The rise of social networking enables the development of multilingual Internet-accessible digital documents in several languages.The digital document needs to be evaluated physically through the Cross-Language Text Summarization(CLTS)involved in the disparate and generation of the source documents.Cross-language document processing is involved in the generation of documents from disparate language sources toward targeted documents.The digital documents need to be processed with the contextual semantic data with the decoding scheme.This paper presented a multilingual crosslanguage processing of the documents with the abstractive and summarising of the documents.The proposed model is represented as the Hidden Markov Model LSTM Reinforcement Learning(HMMlstmRL).First,the developed model uses the Hidden Markov model for the computation of keywords in the cross-language words for the clustering.In the second stage,bi-directional long-short-term memory networks are used for key word extraction in the cross-language process.Finally,the proposed HMMlstmRL uses the voting concept in reinforcement learning for the identification and extraction of the keywords.The performance of the proposed HMMlstmRL is 2%better than that of the conventional bi-direction LSTM model.

Analysis of a simplification strategy in a nonhydrostatic model for surface and internal wave problems

This paper examines the simplification strategy of retaining only the nonhydrostatic effect of local acceleration in a three-dimensional fully nonhydrostatic model regarding the submesoscale wave phenomenon in the ocean.Elaborate scale analysis of the vertical component of the Reynold-averaged Navier-Stokes(RANS)equation was performed,confirming the rationalization of this simplification.Then,the simplification was implemented in a RANS equation-based nonhydrostatic model NHWAVE(nonhydrostatic WAVE)to make a simplified nonhydrostatic model.Numerical examples were taken to test its performance,including surface sinusoidal waves propagating on an idealized East China Sea topography,tidally induced internal lee waves and small-scale solitary waves.The results show that in a considerably wide range of nonlinear strengths,the simplified nonhydrostatic model can obtain similar results as those in the fully nonhydrostatic model,even for smaller-scale solitary waves.Nonlinearity influences the applicability of the simplification.The stronger the nonlinearity is,the worse the simplified model describes the nonhydrostatic phenomenon.In general,the simplified nonhydrostatic model can simulate surface waves better than internal waves.Improvement of computational efficiency in the simplified nonhydrostatic model is reasonable,reducing the central processing unit time duration in the fully nonhydrostatic model by 16.4%–20.6%.The specially designed algorithm based on the simplified nonhydrostatic equation can remarkably reduce the computational time.

Numerical Simulation of the Settling Flux of Biodeposition in a Bay with Cage Culture Through Similarity Theory and a Simplified Pollution Source

The settling flux of biodeposition affects the environmental quality of cage culture areas and determines their environmental carrying capacity.Simple and effective simulation of the settling flux of biodeposition is extremely important for determining the spatial distribution of biodeposition.Theoretically,biodeposition in cage culture areas without specific emission rules can be simplified as point source pollution.Fluent is a fluid simulation software that can simulate the dispersion of particulate matter simply and efficiently.Based on the simplification of pollution sources and bays,the settling flux of biodeposition can be easily and effectively simulated by Fluent fluid software.In the present work,the feasibility of this method was evaluated by simulation of the settling flux of biodeposition in Maniao Bay,Hainan Province,China,and 20 sampling sites were selected for determining the settling fluxes.At sampling sites P1,P2,P3,P4,P5,Z1,Z2,Z3,Z4,A1,A2,A3,A4,B1,B2,C1,C2,C3 and C4,the measured settling fluxes of biodeposition were 26.02,15.78,10.77,58.16,6.57,72.17,12.37,12.11,106.64,150.96,22.59,11.41,18.03,7.90,19.23,7.06,11.84,5.19 and 2.57 g d^(−1)m^(−2),respectively.The simulated settling fluxes of biodeposition at the corresponding sites were 16.03,23.98,8.87,46.90,4.52,104.77,16.03,8.35,180.83,213.06,39.10,17.47,20.98,9.78,23.25,7.84,15.90,6.06 and 1.65 g d^(−1)m^(−2),respectively.There was a positive correlation between the simulated settling fluxes and measured ones(R=0.94,P=2.22×10^(−9)<0.05),which implies that the spatial differentiation of biodeposition flux was well simulated.Moreover,the posterior difference ratio of the simulation was 0.38,and the small error probability was 0.94,which means that the simulated results reached an acceptable level from the perspective of relative error.Thus,if nonpoint source pollution is simplified to point source pollution and open waters are simplified based on similarity theory,the setting flux of biodeposition in the open waters can be simply and effectively simulated by the fluid simulation software Fluent.

A Progressive Line Simplification Algorithm

An application of line simplification considering spatial knowledge is described.A method for identifying potential conflict regions,in order to avoid the self_intersection of generalized lines,is also discussed.Furthermore,a new progressive line simplification algorithm is presented.From the view point of spatial cognition,a spatial hierarchical structure is proposed,and its application to construction of spatial knowledge related to a line is explained.

Simplification and Experimental Verification for Temperature and Humidity Field Coupling Model of Conservatory Soil

Based on the surface energy balance model which is widely used abroad, a temperature and humidity field coupling model of conservatory soil without crop vegetation in full illumination was established. Considering the relatively closed environment in conservatory, weak solar radiation and little surface evaporation of soil, the daily variation of water content in different soil layers may be neglected, then the temperature and humidity field coupling model was simplified to a one-dimensional thermal diffusion model. The simplified model and the temperature and humidity field coupling model adopted the same computational method of soil physical parameters and discrete format of heat diffusion differential equations, and were applied to the continuous simulation of temperature field in conservatory soil without crop vegetation in full illumination. Through the comparison between simulation results and experimental data, the precision of the simplified model was verified. The typical rule of soil heat flux variation in a 24 h cycle was also obtained.

A Collaborative Simplification Method for Multiple Coastlines Based on the Hierarchical Triangulation Network Partition

For the current automatic coastline generalization method,only one-line element is considered separately,but the relationship between the nearby elements is not effectively considered.A synergistic simplification method for multiple coastlines based on the hierarchical triangulation network partition(HTNP)is proposed in this paper.Firstly,the constrained Delaunay triangulation is constructed to partition the regions that can be simplified.Then,a hierarchical binary tree model to structure the morphological characteristics of the above several coastlines and the spatial proximity between different coastlines is constructed.Finally,the small curved and curved invisible parts of the coastline are deleted according to the visual constraints,and the narrow part between the coastline itself or the different coastlines is exaggerated appropriately,and the automatic simplification of the coastline is realized.The experimental results show that:①Relationships between the different coastlines are considered,and the shortcomings of considering the coastline separately are overcome;②Under the condition of the multiple coastlines in complex sea areas,the problem of collaborative simplification is solved,the quality of the coastlines are improved obviously,and the proposed method can be applied into more types of coastlines.

Simplification of English Sentence Pattern Structure in the Teaching of Translation and Grammar

In the teaching of translation and grammar,it is essential that the English sentence structure is analyzed and transformation of sentence structures both in English and Chinese in class and after class. However,many kinds of English sentences structures prevent the students who study English from getting the basic idea and concept of a sentence in translation or grammatical analysis. In this paper,the author submitted the basic mould of simplification of English sentence structure,and the students can learn and master the translating technique through the simplified English sentence structure. Thus it helps to improve students' competence of translation.

Complex Model Simplification Algorithm with Improved Error Evaluation

Given the framework of incremental mesh simplification based on edgecollapse, the paper proposes a mesh simplification algorithm using an improved approach formeasuring simplification error. The algorithm uses edge collapse to simplify the trianglemesh andmaintains surface error approximations using c-error for the faces which have changed after edgecollapse and d-crror for the faces which become degenerated after edge collapse. Also, we reportsome results using a variety of computer graphics models, which can show that the algorithm canachieve the desired simplification effect.

The Solid-liquid Transformation of Low-grade Solid Potash Deposit in Dalangtan Basin and the Simplification of the Liquid Phase System

We studied the solid-liquid transformation of low-grade solid potash deposit in Dalangtan Basin and simplified the liquid phase system.We did experiments to optimize conditions of the solid-liquid transformation.The Suitable

Terrain Simplification Research in Augmented Scene Modeling

Augmented reality is the merging of synthetic sensory information into a user's perception of a real environment. As one of the most important tasks in augmented scene modeling, terrain simplification research has gained more and more attention. In this paper, we mainly focus on point selection problem in terrain simplification using triangulated irregular network. Based on the analysis and comparison of traditional importance measures for each input point, we put forward a new importance measure based on local entropy. The results demonstrate that the local entropy criterion has a better performance than any traditional methods. In addition, it can effectively conquer the 'short-sight' problem associated with the traditional methods.

Application of equivalent resistance to simplification of Sutong Bridge piers in tidal river section modeling

This paper describes some details and procedural steps in the equivalent resistance （E-R） method for simplifying the pier group of the Sutong Bridge, which is located on the tidal reach of the lower Yangtze River, in Jiangsu Province. Using a two-dimensional tidal current numerical model, three different models were established： the non-bridge pier model, original bridge pier model, and simplified bridge pier model. The difference in hydrodynamic parameters, including water level, velocity, and diversion ratio, as well as time efficiency between these three models is discussed in detail. The results show that simplifying the pier group using the E-R method influences the water level and velocity near the piers, but has no influence on the diversion ratio of each cross-section of the Xuliujing reach located in the lower Yangtze River. Furthermore, the simplified bridge pier model takes half the calculation time that the original bridge pier model needs. Thus, it is concluded that the E-R method can be use to simplify bridge piers in tidal river section modeling reasonably and efficiently.

An Adaptive SVP Simplification Based on Area Difference

Sound velocity profile(SVP)data is indispensable in the multi-beam data processing.The sampling density is of great importance for SVP to represent the vertical variation of sound velocity accurately and guarantee the accuracy of sound ray tracing(SRT).However,the SVP also affects the SRT efficiency significantly,especially in deep-sea multi-beam sounding data processing.To improve SRT efficiency and ensure SRT accuracy,an adaptive SVP simplification method based on area difference is proposed in this article.Firstly,the relationship between the area difference of the raw SVP and the simplified one and SRT bias is studied,and the relationship model of them is built.Then,by considering the constraint of SRT accuracy,the SVP simplification method and the simplifying SVP procedure SVP are given.Finally,a deep water experiment is conducted to verify the proposed method.Compared to the existing method,the proposed method improves the robustness,feasibility of SVP simplification as well as the accuracy and efficiency of SRT.

THE METHOD OF GRAPHIC SIMPLIFICATION OF AREA FEATURE BOUNDARY WITH RIGHT ANGLES

Some rules of simplification of area feature boundary and the method of acquiring spatial knowledge, such as maintaining area and shape of area feature, are discussed. This paper focuses on the progressive method of graphic simplification of area feature boundary with right angles based on its characteristics.

Cross-Language Transfer Learning-based Lhasa-Tibetan Speech Recognition

As one of Chinese minority languages,Tibetan speech recognition technology was not researched upon as extensively as Chinese and English were until recently.This,along with the relatively small Tibetan corpus,has resulted in an unsatisfying performance of Tibetan speech recognition based on an end-to-end model.This paper aims to achieve an accurate Tibetan speech recognition using a small amount of Tibetan training data.We demonstrate effective methods of Tibetan end-to-end speech recognition via cross-language transfer learning from three aspects:modeling unit selection,transfer learning method,and source language selection.Experimental results show that the Chinese-Tibetan multi-language learning method using multilanguage character set as the modeling unit yields the best performance on Tibetan Character Error Rate(CER)at 27.3%,which is reduced by 26.1%compared to the language-specific model.And our method also achieves the 2.2%higher accuracy using less amount of data compared with the method using Tibetan multi-dialect transfer learning under the same model structure and data set.

The Spatial Characteristics of a Line and Their Application to Line Simplification

This paper discusses the spatial knowledge related to a line ,and the characteristic points of lines is detected.According to the requirements of line generalization,new algorithms for identifying characteristic line points are presented.These characteristic points are used to improve the algorithms of line generalization.An algorithm for identifying bends is shown.In this paper,improved algorithms based on those by Douglas_Peucker,Visvalingam and Whyatt are shown.In this test,the progressive process of line generalization is emphasized.

Query Expansion Using Wikipedia and a Concept Base in Cross-language Information Retrieval

The present paper describes the use of online free language resources for translating and expanding queries in CLIR （cross-language information retrieval）. In a previous study, we proposed method queries that were translated by two machine translation systems on the Language Gridem. The queries were then expanded using an online dictionary to translate compound words or word phrases. A concept base was used to compare back translation words with the original query in order to delete mistranslated words. In order to evaluate the proposed method, we constructed a CLIR system and used the science documents of the NTCIR1 dataset. The proposed method achieved high precision. However~ proper nouns （names of people and places） appear infrequently in science documents. In information retrieval, proper nouns present unique problems. Since proper nouns are usually unknown words, they are difficult to find in monolingual dictionaries, not to mention bilingual dictionaries. Furthermore, the initial query of the user is not always the best description of the desired information. In order to solve this problem, and to create a better query representation, query expansion is often proposed as a solution. Wikipedia was used to translate compound words or word phrases. It was also used to expand queries together with a concept base. The NTCIRI and NTCIR 6 datasets were used to evaluate the proposed method. In the proposed method, the CLIR system was implemented with a high rate of precision. The proposed syst had a higher ranking than the NTCIRI and NTCIR6 participation systems.

Research on Cross-Language Retrieval Using Bilingual Word Vectors in Different Languages

Bilingual word vectors have been exploited a lot in cross-language information retrieval research. However, most of the research is currently focused on similar language pairs. There are very few studies exploring the impact of using bilingual word vectors for cross-language information retrieval in long-distance language pairs. In this paper, it systematically analyzes the retrieval performance of various European languages (English, German, Italian, French, Finnish, Dutch) as well as Asian languages (Chinese, Japanese) in the adhoc task of CLEF 2002–2003 campaign. Genetic proximity was used to visually represent the relationships between languages and compare their crosslingual retrieval performance in various settings. The results show that the differences in language vocabulary would dramatically affect the retrieval performance. At the same time, the term by term translation retrieval method performs slightly better than the simple vector addition retrieval methods. It proves that the translation-based retrieval model can still maintain its advantage under the new semantic scheme.

采用混合策略联合优化的模糊C-均值聚类信息熵点云简化算法

针对传统聚类算法处理点云简化问题时精度低、耗时长且易丢失特征信息等问题,提出了一种基于动态精英自适应混合策略的鹈鹕算法(DEAMPOA)与加权熵法联合优化的模糊C-均值聚类(FCM)信息熵点云简化算法。采用动态自适应种群混合策略,同时融合了精英反向化思路,显著提升了鹈鹕优化算法(POA)的收敛趋势和全局寻优能力,提高了寻找FCM最优聚类中心的成功率;利用DEAMPOA结合加权熵法对FCM进行优化,提高鲁棒性的同时增强了搜索精度,得到较好的聚类结果;在8种UCI标准数据集上与4种算法对比进行聚类性能评估实验,验证了所提方法综合性能优越;将所提方法与信息熵融合,并应用在三维点云KITTI数据集简化中。实验结果表明:与包围框简化法、随机采样简化法和特征选择简化法对比,所提方法全局误差简化前后点集之间平均欧式距离(MED)指标分别降低了2.25%、6.93%、5.74%,点云简化效果最优且运行速度满足要求。

城市实景模型结构化线面特征重构方法

为了改善城市实景模型边缘模糊,提出了一种曲率引导的结构化线面特征重构方法。根据曲率特征将网格分割为平面、可展凹、可展凸以及不可展曲面4类,在平面分割结果内提取平面,在可展凹和可展凸分割结果内提取直线,对过度弯曲的不可展区域进行保留,最终形成包含几何特征的复合网格模型。结果表明,结合曲率信息预先设置几何特征的潜在范围,使得结构化线面特征更可靠,同时保证城市实景中复杂的树结构不被错误地提取为平面。