Computer and Modernization

Product Feature Extraction Based on Improved LDA Topic Model

SHE Wei-jun, LIU Zi-ping, YANG Wei-fang

2016, 0(11): 1-6,57. doi:10.3969/j.issn.1006-2475.2016.11.001

Asbtract ( 281 )

References | Related Articles | Metrics

Aiming at the problems existing in LDA model used to extract product features, a method combined syntactic analysis and topic model, named SA-LDA, is proposed. Firstly, we analyze reviews under products which belong to a category based on syntactic analysis, extract explicit features and cluster them to get feature set and opinion set, and then construct corpus. After that, opinion sentences are extracted to be used for topic clustering, must-link and cannot-link are constructed for guiding the topic learning and each topic corresponds to a specific feature cluster. Experiments show that the performance of the method proposed in this paper is good in explicit features and implicit features, and it not only ensures recall rate, but also improves precision score compared to other methods.


An Improved CHI Text Feature Selection Algorithm

FAN Cun-jia, WANG You-sheng, WANG Yu-ting

2016, 0(11): 7-11,63. doi:10.3969/j.issn.1006-2475.2016.11.002

Asbtract ( 201 )

References | Related Articles | Metrics

In the process of text classification, feature selection algorithm is a greatly important part. CHI statistics is a classical feature selection method, but it has some defects. Aiming at the shortage of CHI statistics algorithm, on the one hand, in order to take into account the document frequency and word frequency of items, word frequency factor and variance among classes were introduced into CHI algorithm. On the other hand, in order to exclude the items which rarely appear in the specified class and largely appear in other classes, and reduce the error of artificially selecting scaling factor, the adaptive scaling factor was introduced into CHI algorithm. The results show that the improved CHI feature selection algorithm is superior to CHI statistics algorithm in the unbalanced corpus.


Distributed CIF Quadtree Indexing Method Based on Hadoop

XU Huan1, FENG Jun1, ZHANG Peng-cheng1, TANG Zhi-xian2, LIU Yi1, CHEN Zhi-fei1, ZHANG Li-xia1

2016, 0(11): 12-19,24. doi:10.3969/j.issn.1006-2475.2016.11.003

Asbtract ( 233 )

References | Related Articles | Metrics

We design some algorithms about parallel index creation, intersection query and regional remove for the rectangle objects, which are suitable for the distributed environment. These algorithms rely on the methods of dividing the data space, as well as the idea of divide-and-conquer. And they are based on the CIF indexing techniques supported by the Hadoop platform and the MapReduce programming model. On this basis, we test the parallel index creation and intersection queriess efficiency by changing the size of data sets of rectangle objects and the number of the map tasks. The experiments results show that using parallel algorithms of the parallel index creation and intersection queries can improve the processing efficiency for large data sets.


Path Planning for Mobile Robot Based on Relaxed Dijkstra Algorithm

PAN Cheng-hao, GUO Min

2016, 0(11): 20-24. doi:10.3969/j.issn.1006-2475.2016.11.004

Asbtract ( 246 )

References | Related Articles | Metrics

Under the premise of using grid method in environment modeling, the problem of real-time path planning for mobile robot in the larger-scale working environment with dense obstacles is difficult to solve. A new method of the global path planning for mobile robot-relaxed Dijkstra algorithm was proposed according to the characteristics of grid-map structure. A potential Manhattan distance filed from the start point to global points was constructed first by four neighborhood searching in linear time. Then, eight neighborhood searching was used to get a collision-free, near-optimal path from the target point to the start point. Last, the results of Matlab simulation experiments proved that the proposed method was ten times faster than the Dijkstra algorithm and A-star 〖JP2〗algorithm that had used heap sort on computing time, and compared with the shortest path, the length of its path had a reasonable error.〖JP〗


Free Trial Strategy of SaaS with Network Externalities

SONG Qian-qian, MIAO Hong, WANG Nian-xin, GE Shi-lun

2016, 0(11): 25-32,68. doi: 10.3969/j.issn.1006-2475.2016.11.005

Asbtract ( 151 )

References | Related Articles | Metrics

The software products with strong network externalities are society-oriented applications and located before or after the supply chain. They are transforming to Software as a Service. And those services are also offered for free trial to ease uers’ uncertainty and help the diffusion of Software as a Service. Taking network externalities into the free trial problem of SaaS with differences in the quality in the duopoly market, this paper studied the free trial problem in the presence of network externalities through the Bertrand competition model. According to the results, the network externality intensity affects the free trial duration and market profit of the low service quality SaaS provider, and also affects the range of consumer surplus and social welfare.


Virtual Network Embedding Based on Node’s Multiple Attributes Ranking

ZHANG Pei-ying1,2

2016, 0(11): 33-37. doi: 10.3969/j.issn.1006-2475.2016.11.006

Asbtract ( 116 )

References | Related Articles | Metrics

To address the problem that the node ranking algorithm only takes node’s computing resources and neighbor’s bandwidth resources into consideration, this paper proposed a novel virtual network embedding algorithm based on node’s multiple attributes ranking. This algorithm takes advantage of node’s multiple attributes to measure the node’s importance, in order to make the adjacent virtual nodes mapping onto the adjacent substrate nodes, to accomplish the node mapping stage through the breadth first search of graph, to finish the link mapping stage using the K-shortest path algorithm. Simulation results show that, the proposed algorithm can improve the acceptance ratio of virtual network requests and revenue/cost (R/C) ratio.


Low-dose CT Image Reconstruction Based on Adaptive Kernel Regression Method and Algebraic Reconstruction Technique

ZHONG Zhi-wei

2016, 0(11): 38-42. doi:10.3969/j.issn.1006-2475.2016.11.007

Asbtract ( 230 )

References | Related Articles | Metrics

To the problem of sparse angular projection data of CT image reconstruction, TV-ART algorithm introduces the gradient sparse prior knowledge of image to algebraic reconstruction, and the local smooth image gets a better reconstruction effect. However, the algorithm generates step effect when the borders are reconstructed, affecting the quality of the reconstruction. Therefore, this paper proposes an adaptive kernel regression function combined with Algebraic Reconstruction Technique reconstruction algorithm (LAKR-ART), it does not produce the step effect on the border reconstruction, and has a better effect to detail reconstruction. Finally we use the shepp-logan CT image and the actual CT image to make the simulation experiment, and compare with ART and TV-ART algorithm. The experimental results show the algorithm is of effectiveness.


Pedestrian Warning System Based on On-board Videos

HU Peng-cheng1, ZHANG Chao1, BAO Bing-ji1, WU Xiao-pei1, WANG Ying-guan2

2016, 0(11): 43-52. doi:10.3969/j.issn.1006-2475.2016.11.008

Asbtract ( 201 )

References | Related Articles | Metrics

Driving safety research is always one of the hot topics in social life, this paper designs and realizes a real-time pedestrian warning system based on aggregate channel features. The system includes pedestrian detection module, regional division module, monocular vision distance measurement module and warning module. The pedestrian detection module uses the method of combing aggregate channel features with cascade Adaboost classifier to construct channel pyramids, as wellas to quickly detect the pedestrian in video and obtain the key information and motion properties. The monocular vision distance measurement module estimates the distance between pedestrian and vehicle. The warning module judges the degree of danger based on the motion information and presents the response type. This paper made experiments with the record videos in urban environment, and it validated the real-time capability and accuracy of the system.


Dynamic Threshold Selection of Local Region Based on Image Partition

ZHU Hai-yang1, XU Gen-jiu2, LI Yuan-chen1, ZHANG Meng-qi2

2016, 0(11): 53-57. doi:10.3969/j.issn.1006-2475.2016.11.009

Asbtract ( 209 )

References | Related Articles | Metrics

Most of the current edge detection methods adopt global threshold selection, which leads to the missing edge of local image. Moreover, the existing local threshold selection based on image partition causes a big difference of local thresholds and makes the edge discontinuity. Thus, in order to solve these problems, a method of local dynamic threshold selection based on image partition is proposed. Entropy is used to block image. The influence of the neighborhood on dynamics is considered to calculate the local threshold and ensure the continuity of local thresholds. Finally, the method is applied to Robert operator and Canny algorithm. The results show that the dynamic threshold selection can improve the detection accuracy effectively and enhance the continuity of edge.


Moving Object Detection Based on Scene Semantic Prior and Global Appearance Consistency

JIAO Yu-qing, WANG Wen-zhong, LUO Bin

2016, 0(11): 58-63. doi: 10.3969/j.issn.1006-2475.2016.11.010

Asbtract ( 164 )

References | Related Articles | Metrics

Moving object detection in dynamic background is a very challenging fundamental problem in video surveillance. This paper presents a robust moving object detection method. First, we develop an effective ViBe algorithm against dynamic background by incorporating the scene prior information that is predefined in initial frame. Then, the global GMM models of foreground objects and background are estimated by foreground and background pixels detected by the improved ViBe algorithm. These GMM models are employed to classify every pixel effectively and remove some of the false results. For further alleviating the effects of noises, the superpixel-based refinement is adopted to obtain the final results. The experimental results on the collected video sequence with strongly dynamic background suggest that the method significantly outperforms other moving object detection methods.


Concurrency Control Algorithms with Balanced B^link-tree Database Index

BAO Bin1， LI Ya-gang2

2016, 0(11): 64-68. doi:10.3969/j.issn.1006-2475.2016.11.011

Asbtract ( 151 )

References | Related Articles | Metrics

Concerning the concurrency control mechanism for multiversion database index based on Blink-tree， a new Blink-tree concurrency control modification algorithm was proposed. The algorithm divides Blink-tree structure modification into several smaller atomic modifications which run concurrently and deadlock-free. The experimental results show that the new algorithm improves concurrency and transaction throughput， and retains consistency and balance of Blink-tree structure.


A Unified Storage Platform for Unstructured Data

SU Jiang-wen1， KONG Xiao-yun2， SONG Li-hua1, ZHANG Yao1

2016, 0(11): 69-73,103. doi: 10.3969/j.issn.1006-2475.2016.11.012

Asbtract ( 134 )

References | Related Articles | Metrics

This paper researched an electric power business unstructured data unified storage platform. The main implementation technologies and solutions of unstructured data storage platform were expatiated, and the evaluation environment meeting the needs of the testing were build. The test results show the feasibility and effectiveness of the storage platform architecture solutions. This platform can meet the upper various business systems needs used in the field of unstructured data, and full serve the “Big-Centralized” mode proposed by the service power company. The technical solution can support distributed storage and optimize the comprehensive transformation of architecture adaptability for subsequent unstructured platform.


Research and Design of Lightweight Mobile RFID Authentication Protocol

WEI Shu-min, ZHANG Yong-hua, SHANG Yu-fang

2016, 0(11): 74-78. doi:10.3969/j.issn.1006-2475.2016.11.013

Asbtract ( 155 )

References | Related Articles | Metrics

In order to solve the security and privacy issues in the mobile radio frequency identification (RFID) system caused by wireless transmission, a lightweight mobile RFID authentication protocol based on pseudo-random function is provided, and mutual certifications between backend server, reader and tags are achieved. The operation of the protocol is mainly concentrated in the background server and the reader, which can effectively control the cost of the tag. Security analysis shows that the protocol can effectively resist the attack of location tracking, counterfeiting, replay and synchronization attack etc, and the security of this protocol is proved by GNY logic.


Hopf Bifurcation of Delayed Computer Virus Propagation Model

SONG Lei1, WANG Chun-lei2

2016, 0(11): 79-82. doi:10.3969/j.issn.1006-2475.2016.11.014

Asbtract ( 161 )

References | Related Articles | Metrics

Considering that anti-virus software needs a period to clean viruses and there is a temporary immunity for the recovered computers, a delayed computer virus propagation model based on the SIQR computer virus propagation model is proposed. Sufficient conditions for local stability of the positive equilibrium and occurrence of a Hopf bifurcation are obtained by regarding the delay as a bifurcation parameter and analyzing the distribution of the roots of the associated characteristic equation. Finally, a numerical simulation example is carried out to support the theoretical predictions. 

Optimizing Wireless Sensor Network Clustering Based on Boundary Node

YUAN Zhao-zheng1, SHAO Xiu-li1, REN Zhi-juan2， GUO Hai-bo2

2016, 0(11): 83-89,103. doi:10.3969/j.issn.1006-2475.2016.11.015

Asbtract ( 182 )

References | Related Articles | Metrics

An optimization method based on boundary node is proposed, which is used to optimize clustering routing protocol. The boundary node is the node locating at the junction of adjacent clusters. Through dynamic allocation of these nodes, clusters load achieves balancing. Because the boundary nodes are far from the cluster center and need to participate in the balanced load, so the boundary nodes will not become the cluster heads, and the energy consumption is less, thus the multi hop routing forwarding algorithm based on boundary node is proposed by using this characteristic. The experimental results show that the optimization method proposed in this paper can balance the network energy consumption and prolong the lifetime of wireless sensor networks.


Autonomic Flight Path Choice of Aerial Node on Sky-ground Network

LIU Bing-rui， LIU Jing-wei， YAN Chu-ping

2016, 0(11): 90-94. doi: 10.3969/j.issn.1006-2475.2016.11.016

Asbtract ( 131 )

References | Related Articles | Metrics

The restrict of terrain condition can be overcome, communication efficiency of the whole network can be increased and the range of the network service can be extended through taking advantage of sky network including wireless network communication device or terminal carried by unmanned aerial platform, but how to intelligently and efficiently finish the task movement is a difficult issue. The study model including an aerial node and some land nodes is set up and an autonomic aerial node flight path choice is put forward: the multiple hovering points in the target area overhead can be autonomously compared and the optimal one can be selected. The research and test result shows that the optimal hovering point can be autonomously selected in the flight path choice and the network communication coverage of the land network can be provided for a short while, moreover, the computing complexity of above algorithm is quite moderate and satisfies the low energy consumption demand.


Multi-GPU Parallel Framework of Deep Convolutional Neural Networks

YANG Ning

2016, 0(11): 95-98. doi:10.3969/j.issn.1006-2475.2016.11.017

Asbtract ( 215 )

References | Related Articles | Metrics

In recent years, deep convolutional neural network is widely used in the fields of image recognition and speech recognition, and achieves good results. Deep convolutional neural networks are the convolutional neural networks with multiple layers, tens of millions of parameters need to be learned, and computational overhead is large, so the training is very time-consuming. In view of this situation, we propose a multi-GPU parallel framework of deep convolutional neural networks, design and implement model parallel engine, relying on the powerful collaborative parallel computing ability of multi-GPU, combined with the parallel characteristics of deep convolutional neural networks in training, to achieve fast and efficient deep convolution neural networks training.


Optimization of Enterprise Information System Menu Using Association

YU Yue-yang, MIAO Hong, GE Shi-lun, WANG Nian-xin

2016, 0(11): 99-103. doi:10.3969/j.issn.1006-2475.2016.11.018

Asbtract ( 176 )

References | Related Articles | Metrics

Since the ground and development of Internet and cloud computing, enterprise information systems can not only focus on their own business, but also need to consider the user’s behavior and experience. Therefore, modularity and BGLL algorithm were used to divide community structure, and took the information system user behavior into account. This paper analyzed the impact on enterprise information system menu structure optimization. And by comparing case, the results show the usability and convenience of the optimized information systems menu structure which considered association.


Energy-saving Supervision System Based on J2EE

DAI Huan1,2, LUO Min1,2, LIU Bo-ping1,2, ZHANG Jun-min1,2

2016, 0(11): 104-108,113. doi: 10.3969/j.issn.1006-2475.2016.11.019

Asbtract ( 147 )

References | Related Articles | Metrics

Aiming at the faster growth of energy consumption and no strong sense of energy-saving in enterprises, an energy-saving supervision system is designed and developed according to the consumption of electricity, water and oil in the enterprise and demands of the management. Based on the mature enterprise application development standards—J2EE, the system uses MVC and B/S to realize collecting data, analyzing data and saving data. The experiment results show that the system can supervise the energy consumption of enterprises and remind the users to save energy.


Application of Improved SAGA Algorithm in Substation Inspection Job Scheduling

XIE Xiao-jun1, ZHUO Wen-he1, HU Peng2

2016, 0(11): 109-113. doi:10.3969/j.issn.1006-2475.2016.11.020

Asbtract ( 206 )

References | Related Articles | Metrics

Against the substation inspection operation scheduling problem under the condition of multi-resource constraints, according to the location of the inspection members, the current mission, the mission details, the tasks to be implemented, inspection equipment, historical inspection records and other factors, we built a mathematical model, and put forward an improved genetic algorithm. The algorithm solves the problem that the traditional genetic algorithm falls into the local optimal solution, and has the characteristics of fast convergence speed. The experimental results show that the SAGA algorithm is superior to GA algorithm to solve substation inspection job scheduling problem, and has higher calculation efficiency.


Application of Memetic Algorithm in Generating Test Paper Intelligently

YI Gui-sheng, HUANG Wen-hua

2016, 0(11): 114-117,121. doi:10.3969/j.issn.1006-2475.2016.11.021

Asbtract ( 195 )

References | Related Articles | Metrics

Memetic algorithm is a metaheuristic search method. It is often used to solve NP problems. In this paper, through the improvement and optimization of the genetic Memetic algorithm, combined with the characteristics of the intelligent test paper generation, a set of complete solution is put forward. The algorithm uses the Memetic algorithm framework; the global search strategy uses genetic algorithm of piecewise real number encoding; crossover and mutation operations are included in. The local search strategy algorithm using simulated annealing algorithm, solves the local optimization problem effectively. Through the comparison experiment of different algorithms, the experimental results show that the Memetic algorithm proposed in this paper can solve the problem of generating test paper quickly and efficiently, at the same time, the algorithm can improve the quality of test paper, and also can reduce the number of iterations and obtain the optimal solution more quickly.


Fault Diagnosis of Transmission Lines in Power System Based on RS and CPN

SONG Yu-qin, LI Ying, DUAN Jun-rui

2016, 0(11): 118-121. doi:10.3969/j.issn.1006-2475.2016.11.022

Asbtract ( 177 )

References | Related Articles | Metrics

According to the characteristics of power system transmission line network model structure, this paper established a power grid fault diagnosis model based on Rough Sets（RS） and Coloured Petri Net（CPN）. Under the premise of keeping the key information, RS can reduce the extracted fault information and obtain the knowledge of the minimal expression. CPN, based on power line junction line analysis function, can build the CPN model of electric power grid fault diagnosis. Examples show the rapidity and simplicity of the diagnosis model, and the combination of two methods not only improves the generality of fault diagnosis but also reduces the complexity of fault diagnosis.

Design of Fiber Grating Sensor Demodulation System Based on FPGA

ZHOU Ming, WU Xiang-nong

2016, 0(11): 122-126. doi:10.3969/j.issn.1006-2475.2016.11.023

Asbtract ( 196 )

References | Related Articles | Metrics

In order to achieve the low-cost, high speed and high precision fiber grating sensing demodulation, the paper designs a fiber grating sensor demodulation system based on FPGA. The system includes the data collection and A/D module, the FIR digital filter module, the data cache module based on SRAM and the display module. The system provides a global system synchronization clock by the FPGA control module for each sub module. Compared with the conventional fiber Bragg grating sensor demodulation device based on FPGA， the required components of design are in the same piece of FPGA development board. It not only reduces costs but also improves the tone of the instrument system integration. By system experimental verification, the fiber grating sensor demodulation system based on FPGA is successful. Last, an application is proposed about the system, it shows the system has certain practical significance.


Table of Content