Loading...
主 管:江西省科学技术厅
主 办:江西省计算机学会
江西省计算中心
编辑出版:《计算机与现代化》编辑部
Toggle navigation
Home
About Journal
Editorial Board
Journal Online
Online First
Current Issue
Archive
Most Read Articles
Most Download Articles
Most Cited Articles
Subscription
FAQ
Self Recommendation
Contact Us
Email Alert
中文
Office Online
Submission or Manuscript
Peer Review
Editor-in-Chief
Office Work
Instruction
More>>
Instruction
Submit Flow
Template
Copyright Agreement
The Author Changes the Application
Highlights
More>>
Links
More>>
Visited
Total visitors:
Visitors of today:
Now online:
Table of Content
11 September 2018, Volume 0 Issue 08
Previous Issue
Next Issue
Optimization of Train Operation Profile Based on Improved Genetic Algorithm
JI Yun-xia, SUN Peng-fei, MAO Chang-hai, WANG Qing-yuan
2018, 0(08): 1. doi:
10.3969/j.issn.1006-2475.2018.08.001
Asbtract
(
160
)
References
|
Related Articles
|
Metrics
The classic genetic algorithm has been used for the optimization of train operation long ago. However, due to the uncertainty of population evolution direction and insufficient local search ability, the rate of convergence is slow and the quality of solution is low. In this paper, an improved genetic algorithm is proposed to study the optimization of train operation profile. The optimization objective is to minimize the energy consumption of train operation. The constraints are transformed into penalty functions, such as traffic safety, punctuality and precise parking etc. In order to accelerate the population convergent rate and improve the solution quality, a new mechanism is designed, which can guide the evolution direction of the population, and the punctuality adjustment and local search are included in the new mechanism. The demonstrations show that the improved genetic algorithm is suitable for train operation profile optimization and can improve the convergence speed effectively. Moreover, it’s result is more energy saving than the classic genetic algorithm and the adaptive genetic algorithm.
#br# Prediction and Analysis of Stock Price Based on GM-RBF Neural Network
LIU Shu-zhong
2018, 0(08): 8. doi:
10.3969/j.issn.1006-2475.2018.08.002
Asbtract
(
112
)
References
|
Related Articles
|
Metrics
Stock price is usually influenced by various factors in the market, and usually shows nonlinearity and uncertainty in price fluctuation. When we solve the stock price prediction problem, the accuracy of the single prediction method is often low due to its limitations. Therefore, in order to obtain more accurate prediction results, it is necessary to combine two or more prediction methods to establish a combination forecasting model. In view of stock price forecasting, the stock price prediction model based on GM-RBF neural network is proposed. The experimental results show that the stock price prediction model of GM-RBF neural network can predict the stock price more accurately and reflect the law of stock price change more objectively.
An Improved Face Detection Algothrim Based on R-FCN
DAI Hai-neng, MAO Yao-bin
2018, 0(08): 12. doi:
10.3969/j.issn.1006-2475.2018.08.003
Asbtract
(
171
)
References
|
Related Articles
|
Metrics
The region-based convolution network has been widely used in object detection, attracting extensive researcher’s interest. Aiming at the problem of face detection, this paper proposes an improved face detection algorithm based on Region-based Fully Convolutional Networks (R-FCN). In order to make the model training more complete, the online hard example mining method is used to relax the constraints of positive and negative samples, which extends the scope of the training set. For the overlapping problem of face targets, a linear non-maxima suppression method is adopted to avoid missing detection of overlapping faces. The experimental results on the face detection database (FDDB) show that the improved R-FCN model has a higher accuracy than the original R-FCN model.
A Fast Clustering Algorithm Based on Cluster-centers and Partition
JU Shu-cun1,2, CHENG Wen-jie1,2, XU Jian-peng2, XU Xiang2, XU Yang2
2018, 0(08): 16. doi:
10.3969/j.issn.1006-2475.2018.08.004
Asbtract
(
137
)
References
|
Related Articles
|
Metrics
The clustering algorithm based on traditional partition needs to give the number of clustering artificially, and due to the rigid partition of the algorithm, it may lead to the segmentation of large or extended clusters, leading to the wrong clustering results. Clustering by density peak is a new clustering algorithm based on density proposed in recent years. The algorithm does not need to specify the number of clusters in advance, and can detect nonspherical clusters. A fast clustering algorithm based on density peak and partition (DDBSCAN) is proposed in this paper. The algorithm first obtains the cluster center (density peak) of a group of clusters, which describes the “skeleton” of the cluster, then divides the surrounding points into the nearest core object, and finally the clusters is merged by judging the density at the dividing edge. Experiments show that the algorithm can effectively adapt to data sets of arbitrary shape and size, and converges faster than traditional clustering algorithms based on density.
Review of Chinese Entity Relation Extraction
WU Wen-ya, CHEN Yu-feng, XU Jin-an, ZHANG Yu-jie
2018, 0(08): 21. doi:
10.3969/j.issn.1006-2475.2018.08.005
Asbtract
(
527
)
References
|
Related Articles
|
Metrics
Entity relation extraction is an important sub-task of information extraction. It is of great significance for the construction of semantic knowledge base and the development of knowledge graph. For Chinese, semantic relations are more complex, and the effect of entity relation extraction is more significant. So discussing the details of Chinese entity relation extraction methods is very necessary. From the beginning of the emergence and development of entity relation extraction, the current status of Chinese entity relation extraction technology is discussed. Relation extraction methods can be divided into four categories according to the degree of dependence on the corpus: entity supervised relation extraction, unsupervised relation extraction, semi-supervised relation extraction and open domain relation extraction. This paper analyzes and compares these four methods. Finally, the application results and development prospects of deep learning in Chinese entity relation extraction are introduced.
Modified Bat Algorithm Based on WS Small-world Model
YANG Xiao-qin
2018, 0(08): 28. doi:
10.3969/j.issn.1006-2475.2018.08.006
Asbtract
(
139
)
References
|
Related Articles
|
Metrics
Bat algorithm is a new heuristic algorithm based on the observation and study of microbat echo,which is inspired by finding the relationship of bat echolocation behavior and the optimization objective function. Although the bat algorithm has powerful search performance, but it has a relatively simple local search mode and lacks of information sharing between individuals lead to bad search performance. Despite there are many improved algorithms proposed until now, few modified algorithms focus on high-dimensional optimization. Considering individuals have close relationship, which is complex network structure. The relationship is similar to small-world model. Therefore, WS small-world model is employed to optimize bat algorithm. Dynamic neighborhood structure is generated using the feature of WS small-world model edge breaking reconnection, which can improve the overall search capability. An example shows that the general bat algorithm can be used for local search.
Jump and Correction of Fresnel Diffraction Phase Calculated by FFT
XIANG Hong-li1, FAN Qi2, LI Yun3, WANG Yun-fei2
2018, 0(08): 35. doi:
10.3969/j.issn.1006-2475.2018.08.007
Asbtract
(
257
)
References
|
Related Articles
|
Metrics
When the phase of Fresnel diffraction is calculated by Fourier transform algorithm, the phase distribution curve of the receiving surface will be hopping when the phase is wrapped, and if it is unwrapped, it will inevitably lead to the wrong result. The reason that the phase jump on the receiving surface is caused by the Fourier transform algorithm is analyzed. It is because the FFT’s way of indexing the matrix is different from that of the DFT, which leads to the difference between the phase and the real phase of the calculated result. The phase jump can be corrected by doing fftshift respectively before and after the FFT operation. The Fresnel diffraction integral of the two-dimensional moment holes is simulated by using a single FFT, and the phase jump of the receiving surface is corrected by using two times cepstrum, which proves the feasibility of the proposed method.
Kinetic Simulation and Analysis of Normal Human Gait
XU Zhong-hua, FANG Juan,CHEN Long-fei, MU Zai-le
2018, 0(08): 39. doi:
10.3969/j.issn.1006-2475.2018.08.008
Asbtract
(
131
)
References
|
Related Articles
|
Metrics
In order to study the normal human gait, the kinetic simulation and analysis of the lower limb are carried out by Simulink/SimMechanics. Firstly, the kinetic simulation model is built based on the two-linkage mechanism, and the model parameters are set. The results of the simulation are consistent with the inverse dynamic results obtained by Lagrange equation, which validates the correctness of the model. Then based on the two-linkage model, a model of an unilateral lower limb under normal gait is established. And the effectiveness of the model and the influence of ground reaction force on the normal gait is analyzed. The experimental results show that the total torque area of the lower limb under normal force simulation is 23% of the total torque area of overground experiment, which proves that the model is effective. And the total torque area deviation obtained from the simulation with and without ground force reaction is 89%, indicating that the ground reaction has important influence on the normal walking gait.
Annotation Processing Technique of Point Features Based on Vector Tiles
QI Ya-guang1, HU Ming-xiao2, GONG Zhi-hong3, FAN Bing-jun 3
2018, 0(08): 44. doi:
10.3969/j.issn.1006-2475.2018.08.009
Asbtract
(
156
)
References
|
Related Articles
|
Metrics
The vector tile is small in size and highly compressible, which is less constrained by the network bandwidth overhead and storage space. Compared with the desktop point annotation processing, Vector tiles not only faced the problems that the important annotations are overlaped by the minor annotations, the same level annotations are overlaped mutally annotations and elements are overlaped each other, but the problem that the annotation is cutting by the edge of the tile is not fully displayed. These problems have seriously affected the readability of the map and the function of information transmission. Aiming at the above problems, this paper summarizes the principles of vector tile point annotation processing, designs the organization structure of vector tiles, determines the JSON organization form of annotation collocation table, clears the drawing process of vector tile point annotation, uses quadtree coding to target filtering according to the characteristics of vector tiles, and uses R-tree as an efficient spatial index, and uses note-based priority obstacle avoidance technology to solve the above-mentioned problems caused by processing points.
Location and Identification of Navigation Lights and Light-emitting
MA Yun-long1, GAO Yun-ling1, QING Li-fei1, XIONG You-ran2
2018, 0(08): 51. doi:
10.3969/j.issn.1006-2475.2018.08.010
Asbtract
(
107
)
References
|
Related Articles
|
Metrics
In order to realize the automatic cleaning of navigation lights, a method based on monocular vision is proposed to locate and identify the navigation lighting and its light-emitting ports. By extracting the outline of the lamp in the camera image, the least squares method is used to fit the ellipse model of the lamp without lighting, and the pixel coordinates of the center of the lamp are obtained. The centroid method is used to calculate the pixel coordinates of the light source center in the light emitting state. Through the calibration of the camera, the ground constraint coordinate system is established to achieve the conversion from pixel coordinates to world coordinates, and the relative position between the camera and lamps is obtained.
An Improved Canny Image Segmentation Algorithm
YANG Shao-ling, DIAO Yan, LUO Hua, XU Tian-xiong
2018, 0(08): 57. doi:
10.3969/j.issn.1006-2475.2018.08.011
Asbtract
(
126
)
References
|
Related Articles
|
Metrics
Aiming at the shortcomings of image segmentation algorithm, such as poor noise robustness, small boundary information easily loss and narrow application scopes, we improve the threshold value of Canny edge extraction algorithm and extract weighted fusion with chromatic feature. According to the adaptive problem of the Canny operator threshold, the threshold is determined by calculating the class variance between the image background and the target to reduce the error probability. Then, we propose three color features of R, G and B on colorful images with rich information. The new segmentation images are fused by thresholding and segmented images extracted by chromaticity feature extraction. The algorithm not only overcomes the problem of edge information loss and poor robustness of the traditional segmentation and extraction algorithm, but also improves the unit accuracy of the detail. The experimental results show the effectiveness of the improved Canny edge algorithm.
Coupling Sample Prior Distribution Weighted Extreme Learning Machine
XI Xiao-yan, YU Hua-long
2018, 0(08): 61. doi:
10.3969/j.issn.1006-2475.2018.08.012
Asbtract
(
148
)
References
|
Related Articles
|
Metrics
Extreme learning machine can be widely used in classification, clustering, regression, etc. However, previous researchers ignore the influence of sample prior distribution information for classification performance when they deal with class imbalance problems. Aiming at this problem, this paper presents an algorithm called CPWELM ( Coupling sample Prior distribution Weighted Extreme Learning Machine), which is based on extreme learning machine. We fully discuss the importance of the different distribution sample points, then we construct the cost matrix with it for the improvement of classifier performance. We do experiments on 12 imbalanced datasets to verify the feasibility and effectiveness of the proposed algorithm. The results indicate that the proposed algorithm generally performs better than the state-of-the-art ones.
Data Analysis Method of Campus Profile Based on University Website
WANG Song-song, GAO Wei-xun
2018, 0(08): 66. doi:
10.3969/j.issn.1006-2475.2018.08.013
Asbtract
(
142
)
References
|
Related Articles
|
Metrics
This paper mines and analyzes university official website data, and proposes a phrase similarity calculation method based on the combination of phrase tree structure and CilinSimHash algorithm. The algorithm first converts the phrase into a tree with numbers as the root node to calculate the similarity, and then the Tongyici Cilin and SimHash algorithms are combined to calculate the similarity based on CilinSimHash algorithm, finally the similarity method based on phrase structure is weighted with the similarity method based on CilinSimHash algorithm to achieve the phrase similarity calculation. The algorithm is applied to the process of data analysis of university official website, and then the cluster analysis of university official website data is studied, the relationship between university official website data and college evaluation index is achieved. According to the structured data obtained from the official website data of colleges and universities, the clustering algorithm is used to analyze the related index data, which shows the unbalanced development of higher educations at different educational levels.
CART-based Risk Assessment of Community Correction Staff
WANG Yin, GUO Hong-yu
2018, 0(08): 73. doi:
10.3969/j.issn.1006-2475.2018.08.014
Asbtract
(
133
)
References
|
Related Articles
|
Metrics
The researches on the risk assessment of community correction staffs stay in the stage of theoretical at present. Most of models are lack of the support of actual data. For easy to master the actual situation of community correction, based on the CART (Classification and Regression Trees) algorithm, this article analyzes the community correction staff data (excluding sensitive information) for modeling. Besides, via the basic characteristics of prisoners, we get the prisoners risk assessment results, and the accuracy is 85% to 90% compared to the actual results. In addition, this paper discusses the lack of the data in the community correction and proposes several solutions.
Range Pattern Match Query Based on Spatial-temporal Label Trajectories
LIANG Jun-xiu, XU Jian-qiu
2018, 0(08): 79. doi:
10.3969/j.issn.1006-2475.2018.08.015
Asbtract
(
222
)
References
|
Related Articles
|
Metrics
According to semantic descriptions of spatial-temporal label trajectories, combined with traditional moving object queries, the range pattern match query is introduced with formal representations. The range pattern match query returns all the trajectories that match a given query pattern within a given spatial-temporal range, the range pattern match query algorithm based on LR-Tree is designed, and this paper analyzes the filter and the refine progress of the query algorithm. Through extensive experiments with different parameters of the query algorithm, this paper compares with the query algorithms based on RR-Tree, 3DR-Tree, TB-Tree and SETI, and verifies the efficiency of the proposed algorithm.
Research and Application of Big Data Complex Event Pattern
ZHAO Hui-qun, QIAO Yu-heng
2018, 0(08): 86. doi:
10.3969/j.issn.1006-2475.2018.08.016
Asbtract
(
126
)
References
|
Related Articles
|
Metrics
Complex Event Processing (CEP) is a technical method that comes with streaming data. It is used to find interesting event patterns in event streams that are mixed in different data sources. However, with the increasing amount of data, the traditional CEP techniques often fail to address the processing needs for efficient access to event patterns on big data sets. In response to this problem, this paper combines the idea of cluster analysis and association rules in data mining, proposes a “complex event processing” algorithm, and deploys it to the distributed platform Hadoop, thereby discovering the relationship between complex events in 〖JP2〗big data sets and effectively changing the limitations of traditional technologies facing massive data. Finally, the algorithm is applied to GPS big data set and the complex event patterns are found out. Experiments show that the method is feasible and effective.
Prediction of Comprehensive Strength of World Cup #br# Team by Multivariate Analysis Algorithm
LAI Cong-lin, LI Li-ka, ZHANG Hui-chang
2018, 0(08): 92. doi:
10.3969/j.issn.1006-2475.2018.08.017
Asbtract
(
222
)
References
|
Related Articles
|
Metrics
World Cup as an international event is increasingly attracting the attention of the public, the team strength and performance evaluation are always a hot topic. Football tactics performance analysis shows that the match results are closely related to the teams’ statistical indicators, which make the objective and scientific evaluation of the team comprehensive strength become feasible. In this research, clustering analysis and Principal Component Analysis (PCA) methods are used to evaluate the comprehensive strength of the World Cup teams. On the basis of the statistical indicators and the results of 2010 World Cup, a model is built to find out which are the most influential statistical indicators. In addition, the statistical indicators and the results of 2014 World Cup in Brazil are used to verify the model. The experimental results show that the average interception, average shooting number, ball possession number and long success rate are the most influential statistical indicators to reflect the team strength. To a certain extent, the statistical indicators of the group match can predict the promotion results.
Java Web Application Based on Spring MVC Framework
GE Meng, HUANG Su-ping, OUYANG Hong-ji
2018, 0(08): 97. doi:
10.3969/j.issn.1006-2475.2018.08.018
Asbtract
(
1136
)
References
|
Related Articles
|
Metrics
The Web framework based on MVC mode can separate the view, model, and controller of the application, and can simplify the implementation of the control layer. This paper proposes the development method of Java Web application based on Spring MVC framework. First, it analyzes the core components of the Spring MVC framework and the interaction process between them. Second, the basic application of the Spring MVC framework is expounded from two aspects of the configuration file and the implementation of Controller component. Finally, the solution of Spring MVC is given for the three core problems of Web application which are exception handling, interceptor and data validation, Taking the target assessment management system as an example, a part of the implementation code is given. The experiment shows that the Spring MVC framework can improve the stability, scalability and maintainability of Java Web applications.
Importance Analysis for Android Permission Groups
BAI Jun-ze, YANG Hong-li, ZHANG Biao
2018, 0(08): 102. doi:
10.3969/j.issn.1006-2475.2018.08.019
Asbtract
(
104
)
References
|
Related Articles
|
Metrics
The runtime permission model of Android system proposes a new concept of permission groups. As application markets lacking description information of permissions, users are unable to know the importance of permission groups in the application and its setups. Aiming at this problem, this paper proposes an importance analysis method for permission groups, which analyzes an application through reverse engineering, ranks permission groups with the help of ranking algorithm, determines the usages and setups of permission groups in the application, and uses machine learning to test and evaluate this analysis method. The experimental results show that the proposed method can analyze importance of permission groups in the application and recommend appropriate setups for permission groups.
Time Series Analysis of Hot Event Based on Association Rules
WANG Yi-wen, LIU Xin, CAO Shuai, WANG Feng
2018, 0(08): 108. doi:
10.3969/j.issn.1006-2475.2018.08.020
Asbtract
(
150
)
References
|
Related Articles
|
Metrics
As the hot event includes a number of related topics in the development process, analyzing the evolution and propagate path in time-sequence of topics can deeper grasp the specific details of the emergence, development and demise of hot event. In this situation, a method of time series analysis of hot event based on association rules is proposed. Firstly, the frequent keyword sets of multiple time slices are obtained by implementing association rules in parallel. Secondly, the association rule sets are obtained by selecting all of association rule of frequent keyword sets, so as to get keyword sets of multiple topics. Finally, the evolution and propagation path of multiple topics are analyzed according to the keyword sets. Experiments show that the method can track the dynamic change process of hot events comprehensively and effectively, which provides reference and support for network public opinion on the monitoring and management.
Knowledge Graph Construction Method for Carbon Trading
WANG Liang-yu
2018, 0(08): 114. doi:
10.3969/j.issn.1006-2475.2018.08.021
Asbtract
(
626
)
References
|
Related Articles
|
Metrics
In order to solve the problem of data integration in carbon trading, a new method of constructing carbon trading knowledge graph is proposed. For processing the semi-structured and unstructured data in carbon trading, a self-defined Web data wrapper and a method by combining BiLSTM-CRF model with dependency parser are applied to extract triples from data respectively. Then a complete carbon trading knowledge graph can be obtained by transforming the acquired knowledge into linked data, and the semantic query of it can be achieved by fuseki based on Jena. The experimental results show that the proposed method can construct carbon trading knowledge graph rapidly and effectively, and can retrieve useful information from the massive data of carbon trading.
#br# Teaching Service Management System #br# Based on Feature Variability Modeling
ZHENG Xiao-juan, LIU Yang, CHEN Xiang-ke
2018, 0(08): 120. doi:
10.3969/j.issn.1006-2475.2018.08.022
Asbtract
(
121
)
References
|
Related Articles
|
Metrics
The modeling of variability is a hot research topic in the field of software product line. At present, the research is limited to the demand stage, which lacks a complete theoretical system and detailed guidance from field engineering to application engineering. Aiming at this problem, this paper improves the existing method and process of constructing feature models to support the software life cycle, strengthen the mapping between the various models, enhance process operability, and ensure consistency between models. Finally, the demand phase variability model and the design phase variability model are obtained, which are successfully applied to the secondary development of teaching service management system. Through component development and efficiency comparison, it is verified that the improved method provides effective support for reuse.