Search

In this paper, high performance real-time detecting and tracking technology for news events are proposed. The difference to previous TDT articles in the detection and tracking is that this attempt to establish the historical hotspot events Corpus first, and analyze the characteristics of the corpus according to timeline. HMM-based named entity recognition model is also used to find out other event characteristics except those reference timeline. By combining the above two methods and SVM,this study proposes its models in detecting and tracking. Experiments presented in this paper shows the models having a high performance in recall and precision.

With the development of the society and the progress of the science and technology, the demands of the customers are getting higher to the reliability of distribution. On the basis of analyzing the network topology of the distribution information system, this paper designs the integrated mode based on C/S and B/S, analyzes the work flow, and describes in detail the main function modules and the key technologies including the release of spatial data and the simulated algorithm of distribution. In this paper, the logic model of the Distribution Information Management System (DIMS) and the database are also designed. DIMS is developed with WebGIS (Web Geographic Information system) middleware and MapInfo OLE+ embedded techniques. It is practically proved that DIS can meet the real-time requirement of the multi-users and implementation functions such as remote power line simulation, output of drawing, and the control of spatial data version.

In this paper, a type of improved hybrid strategy is used to carry out the image segmentation of gold immunochromatographic test strip, which combines the genetic simulated annealing algorithms (GSAAs) and error back propagation (EBP) neural network algorithm. This hybrid strategy can work out these problems in image segmentation of gold immunochromatographic test strip: the area of test strip is pretty small; the breadth of test line testing zone, control line testing zone will not be fixed due to the reaction between samples and test strip. Computer simulation on this hybrid strategy is realized by Visual C++. The results of computer simulation demonstrate that, comparing with the EBP algorithm, the convergence precision and training speed of the EBP neural network can be improved due to the parameters produced from GSAA. Furthermore, comparing with EBP algorithm, the convergence precision and convergence speed of GSAA is much better. At the end of this paper, this hybrid strategy is applied to image segmentation of test strip and get satisfactory effect.

This paper reviews the Internet of Things and its integral part of the structure based on RFID technology and discusses the status of their application. For the quality control problem in the coal transportation, an idea of using the RFID technology in coal quality monitoring and tracking system has been put forward. The paper also discusses the application framework.

E-Government presents a new and innovative approach to addressing traditional problems of government services utilizing the Internet and World-Wide-Web. After introducing the meanings and benefits of E-government, the paper discusses China's current e-government information security challenges, which cover problems of information, network, equipment and management. Besides, this paper presents effective recommendations to overcome them to cope with these challenges of e-government information security from the technical, managerial and cultural perspective respectively.

Bayesian Network is one of the most effective theoretical models for uncertainty knowledge expression and resoning. In the paper, we simply discuss how to diagnose network fault using Bayesian network based on inference algorithm. This paper illustrates the construction of bayesian network model for system and basic inference model of bayesian network and instance. This paper raises an instance using inference model and illustrates inference algorithm based on bayesian network and emulation.

The paper analyzes the security of conventional network access authentication technology and provides a novel IP address pair based short signature technology which strengthens the security of authentication and control. Conforming to TNC architecture, the paper extends the endpoint access domination to MPLS backbone by IP address pair based short signature technology and depicts a new trustworthy network access scheme and realizes it on Network Processor. The scheme enriches the integrity measure of security client and administrator's security policy with ARP virus, fraud DHCP server, address spoofing of IP and MAC, thus enhances the safety and initiative defense ability of enterprise and public networks, and reduces the cost of enterprise for security deployments greatly while achieving the SP's excellent Return-on-Investment.

Data normalization is the first step of pattern recognition (PR). Feature selection is another part of work in PR. All of them are very important and challengeable. In this paper, some normalization methods and some feature selection methods are introduced first, then, discussed the relationship between feature selection and data normalization. Finally, an algorithm is described how to normalize supervised multi-class (phenotype) data that all the classes are the type of multi-sample by a statistical way. By the method, on the training phase of PR, useful features can be obtained with the same normalization algorithm. The method in this paper has practical significance.

Grid intelligence measure-control is an emerging trend which regard sensor as self-governed networked node. This system required that the sensors have some function of grid and intelligent Intelligence. This paper shows how to use the embedded service technology, ARM9 microprocessor, embedded Linux,Intelligence agent and embedded service toolkit gSOAP implement intelligence and grid function of sensor. In the paper's end, the testing result of the sensor system was also given.

In order to improve the traditional mechanical fault diagnosis expert system, case-based reasoning (CBR) is introduced in this paper for fault diagnosis and a CBR fault diagnosis model is established. In the process of case retrieval, an improved grey correlation model is used to calculate similarity between cases. This model can accurately retrieve similar cases and improve the speed of case retrieval. Finally, a bearing fault diagnosis expert system is established with the method proposed in this paper.

With the development of embedded Real- Time Operation System(RTOS), researches on how to improve the performance and efficiency of RTOS are hot nowadays. Big programs are often composed of several traffic modules while the communication between them has close concern with the system efficiency. In this paper, a new communication model for multi-tasks in VxWorks operation system is proposed which is based on UDP transmit mode and concept of virtual device. It overcomes the shortages of traditional communication theory and satisfies the real-time performance as well as system resource requirement. The paper analyses the principle, possibility and superiority of the model.

In this paper, Computer Generated Forces (CGF) behavior modeling was studied from the viewpoint of associate data mining, for the large quantity of data, rules and models in its process. Because CGF behavior models data source was the combination of staticDB and dynamic data stream, the paper advanced the methods of item truncation and aim-pattern restriction. Through pretreatment, coding, searching frequent pattern, generating associate rules of the CGF behavior modeling data, then decision could be made according as these rules. Application of the two methods improves on the classical aprior algorithm, also improves efficiency of searching frequent items and credibility of CGF's decision. Finally, the application of associate rules mining in air-combat is studied in detail. As the simulation shows, comparing with the traditional matching-rule decision, associate rule mining has higher efficiency on condition with guaranteeing reliability of decision.

This paper propose a VPN scheme based Trusted Computing Platform with discussing the two main technologies used to implement them today: IPSEC and SSL. This paper's main idea is according user's situation, choosing the appropriate VPN authentication pattern,and improving the security of the terminal user with the Trusted Computing Platform (TCP). Finally, give the authentication scheme in the client to site pattern and the security analysis.

This paper introduces five methods of combined programming base on VB and Matlab, and mainly introduces the ActiveX OLE method that is the most popular at present by its application in mining area land information system. This method has many advantages such as high automatization and efficiency, occupying less system resources and strong data exchange ability. In the paper, some basic codes are shown, which interested readers can expand then according to their needs.

The effectiveness in communication and the great economization in time and cost make the video conference system become an important application today. If the video conference system can be constructed on Internet, its practicability and application value will be improved greatly. However, with traditional web page techniques, it is difficult to construct a video conference system that has many complex functions. In this paper, we study the method of developing on-line video conference system with the RIA (Rich Internet Application) technology, and build a prototype system by using Adobe Flex and FMS (Flash Media Server) as platform, which are two important tools in RIA. In the system, multi-users can login the conference page and communicate with each other by video and voice in real-time. The system also provides text chat function and certain administrator function. Experiment results show that our system is feasible and effective. The technique provided in this paper is also a good reference to develop other on-line video communication applications and other rich Internet applications.

In this paper the conception, feature and Technical Composition of smart grid is expounded, and the development process and present research status of smart grid home and abroad as well as the practical significance of developing smart grid in China are introduced. The paper analyzes the characteristics and direction of China's smart grid development, and point out that the development of smart grid requires the development of many smart grid technologies and the building of grid structure, and detailed describes the realization of the self-healing function as well as distributed power generation technology.

The discovery of underlying community structures is a common challenge in the analysis of network data. This paper presents a novel algorithm which gives consideration to both modularity and robustness of the clustering. It's based on SCAN which can detect more details about clusters, hubs and outliers. In order to demonstrate the advantages of this new algorithm in coordinating networks scale with the same similarity, an evaluation is implemented by the classical karate datasets at the end of the paper.

This paper focuses on the issues on massive data compress such as massive data storage, complexity and variability in information processing system (IPS). In order to tackle these issues, this paper proposes an algorithm using multi-decomposition wavelet change for massive data compress (MDC). This algorithm concentrates on the multi-decomposition of data flow, vacuum, and parameters to form a data flow according to the date firstly. Secondly, multi-reversion methodology is used to convert the massive data flow. Finally, an experiment is executed to test this algorithm. The experiments illustrates that this algorithm using wavelet can achieve excellent compress efficiency and time cost while data searching under limited vacuum.

Intrusion detects is an important method to ensure the network security and the distributed intrusion detect system can detect the intrusion of the entire network. The association analysis is practical and feasible to improve the detective performance of intrusion detection system. The paper proposes a tree-layer alert data correlation analysis model of distributed intrusion system to reduce the false alerts by analyzing the intensity of the alert data, clear or reduce the repeated alert by clustering the alert data and discover the high-level attack tactics by associating the alert data. The paper provides the algorithm of each module, and the experiment of the high-level event correlation module with the detect attack data Mitnick shows that the association based on the ontology can achieve the detection of the process of the multi-step distributed attack.

Attribute importance ranking still is a key problem required to be solved for the classification problem. The lack of efficient heuristic information is the fundamental reason that affects the attribute selection in data mining. In this paper, to determining the importance level of the attributes, a new measure based on partial derivative distribution of the classification hypersurface output corresponding to the input attributes is proposed. This paper indicates that if the classification hypersurface is acquired by SVM, it is more convenient to measure the attribute importance ranking. The validity of the proposed method is experimentally evaluated, the experimental results prove that the approach is more efficient.

In this paper, the mathematic model of the optimum design parameter for automated pharmacy was put forward, and used a comprehensive learning multi-objective particle swarm algorithm optimize. The simulation experiment is given, which aimed at actual pharmacy case. The result of this paper provides that the algorithm can figure out multi-objective for the automated storage and retrieval system of the pharmacy.

Currently, most mobile learning platforms are heterogeneous so that they always serve to specific types of mobile devices, and other mobile devices cannot access to them. The paper presents the problem and makes a research on combining mobile learning with eLML. ELML, which is released by University of Zurich (Switzerland), is a kind of e-Lesson markup language based on XML framework. It is independent on any special learning platforms and abides by SCORM standard. Additionally, eLML has features of being powerful transplantable capability and easily integrates with mainstream e-learning platforms. Therefore, it recommends applying eLML to the mobile learning application field. However, there are two problems, which have prevented eLML from using at mobile learning platforms. (1) Not all mobile devices can read eLML contents directly; (2) eLML cannot recognize mobile devices automatically and cannot optimize e-Lesson contents in accordance with capabilities of mobile devices. To solve them, the paper proposes to make use of multi-serving technologies to recognize mobile devices and then generate mobile learning materials according to devices' capabilities. On the above of studies, the paper introduces an approach on how to design and implement a mobile learning system architecture based on eLML framework. It also finds a solution on how to integrate eLML mobile learning materials with current e-Learning platform seamlessly.

In this paper, based on previous technical experience gained by developing ROST Content Mining Software and ROST Emotion Analysis Software, the author proposes a model system which extracts and evaluates the information obtained from the combination scan of multi-sentences, single sentence, phrase and single word for opinion mining and sentiment analysis. In particular, the paper introduces the system's overall structure, its implementation process and further describes the key technology involved in detail. Meanwhile, the principle of the key technology and its core code are also presented.

Checkpoint is very important to system fault tolerance; the user space memory of specified process is saved as completely as possible in both kernel-level and user-level check pointing. While user space memory on Windows platform is divided into several blocks, which are possible in different states, but access state (i.e. committed state) is only one of them. This paper discusses the principle of user space memory management on Windows platform, and especially reveals the mechanism of region which includes virtual address continuous pages with same attributes. This paper presents the structure of user space virtual memory and the approaches to saving and recovery of user space virtual memory. The result is the same as expected.

With the rapid development of information technology, multi-sensor information fusion research has been made great progress, currently there are Bayesian test, the voting method, fuzzy sets, neural networks and other methods algorithm being used in multi-sensor information fusion commonly. However, there are many shortcomings which are difficult to overcome in the process of fusion by these methods, so now the international community has already proposed to apply the D-S evidence theory to information fusion. This method does not require the priori probability, and it's very useful when it is used in information fusion, but the information with the conflict can not be effectively fused by using D-S evidence theory. Larger conflict in the evidence, the fusion result is more easily prone to false consensus. Therefore, this paper presents an improved information fusion method based on original D-S evidence theory, and this method can deal with the evidences with conflict effectively. In this paper, the improvement is in the denominator of the D-S evidence theory. When the evidence is purely of conflict, the denominator is zero, the formula does not make sense. This improvement thinking is coupled with sum between the orthogonal sum of all of evidence and product, which is between the conflict factor and the inverse value of the standard deviation of the support degree, and then it subtract maximum difference of the basic probability of the evidence. If the conflict of the evidence is greater, this advantage of this method is more obvious. If there is no conflict in the evidence, the fusion result is consistent with item of original D-S evidence theory. Finally the simulation experiment gives a set of data with more conflict. The comparative data from the experiments show that the improved method of information fusion is essential to improved in resolving conflict issues, and it is effective.

XML-SNMP based network management model is essential parts of the network management infrastructure. It is generally recognised that there are two problems remained unsolved for this kind of model. These two problems are load balance of the XML manager and the potential bottleneck of the XML/SNMP gateway. This paper presents a novel XML-template based approach to tackle these two problems. We developed three XML templates including host, graph, and data query templates. This paper gives detail discussion of the three templates in regard of the structure, content and relationship as well as the mapping mechanism in the relational databases. The experimental results performed on a real deployed XML-SNMP system demonstrate the effectiveness of the proposed approach. When compared with traditional XML-SNMP system, our approach effectively reduces the types and quantities of management data and overhead of translation between XML and SNMP. The proposed approach solves the load balance and bottleneck issues resulting in significantly overall performance improvement of the XML-SNMP model.

This paper proposes an exception handling mechanism for developing reliable smart process-based applications. With Smart Process-based Application Model that we have proposed, a smart process is generated automatically according to the execution of composite web services to better meet user's individual requirements. We focus on the failures of a smart process and give an approach to handle exceptions. Firstly, based on Execution Transition Diagram(ETD) and several execution states we define in this paper, exceptions can be handled automatically by selecting rational transitions in the ETD during the execution of a smart process. And then, with Smart Process Logic Model(SPLM) and Withdrawal Business Spanning Graph(WBSG) we have proposed, business services that need to be withdrawn could be achieved automatically.

E-mail is the most commonly used network application which has become an important way of network communication. The research of content-based spam filtering technology is a key problem of the Internet security field. The paper described a spam filtering which was based on the Naive Bayesian algorithm and designed a spam filtering model based on the naive Bayesian algorithm. The paper carried out an experiment on the Ling-spam Corpus with VC 6.0 platform, and got a higher accuracy and recall rate.

After analyzing the basic principles of the spread of USB flash disk viruses, this paper concludes that the spread of USB flash disk viruses need two necessary conditions, and on the basis, proposes two counter measures which can prevent the spread of USB flash disk viruses thoroughly. At last, this paper proposes two designs to prevent the spread of USB flash disk viruses.

In this paper, we revisit the frequent itemsets mining algorithm F-Miner which is an efficient algorithm on both sparse and dense databases. This paper focuses on how to reduce the time consumption of the outputting stage. This paper proposes two efficient output optimization strategies, output an item as string and compress the frequent itemset. Using these two strategies, we can improve the performance of the algorithm further. Experimental results show that the two output optimization strategies are efficient on both sparse and dense databases.

Software evolution is characterized by its changing in accord with environment and requirement. It is one of the most difficult and important problems of software. From the runtime perspective, it includes two types: static evolution and dynamic evolution. Static evolution is mainly implemented by human static intervention and analysis. Dynamic evolution is based on the runtime evolution of software, mainly characterized as software self-adapting to environment. This paper presents a change-tracing model that regards the change as the starting point and focuses on software evolution in OO paradigm. It illustrates the change transmission and implementation process during the whole software lifecycle. Concerning the similarity between the change transmission process and the software lifecycle, a research framework of software evolution is proposed. This framework integrates all elements that are involved in the change transmission process, such as scenario, object, component and SA (Software Architecture). Based on this framework, the research of software evolution will be more effective and ties in dynamic evolution. At the end of paper, it gives some detailed discussion of key problems of software evolution in the framework and the corresponding solvability ideas.

As for the limitation of static information verification in the current remote attestation models, this paper presents a novel remote dynamic attestation model named BTRAM (Behavior based Trust Remote Attestation Model) which is based on the user behavior trustworthiness and attribute-based access control. In this model, a verifier proxy is used to verify the requestor. Meanwhile, the executive process and formal description are given in this paper. Finally, a performance analysis is given to the model. Based on the dynamic behavior of the communication entities, the model can not only achieve dynamic authorization, but also overcome the complex management issue of role-based remote attestation. Using a verifier proxy to verify the trustworthiness of requestor can avoid the leakage of requestor-platform's configuration, which may cause an attack to the requestor. Taking user behavior evidence as a trusted evaluation factor makes remote attestation more trustworthy than traditional remote attestation mechanism based on platform identity and integrity.

The realization of the Intelligent Transportation Systems will effectively solve the problem of traffic congestion and urban traffic pollution, improve the road capacity and traffic safety. A crucial key of the realization of the ITS is the estimate and prediction of travel time: how to make and continuously update prediction of travel time for several minutes into the future using real-time data. In order to analyze the trend of the travel time accurately, combined with superiority of wavelet in dealing with time-varying information, a combined method for travel time prediction based on wavelet denoising and exponential smoothing and metabolic curve fitting model is presented in this paper. The new method has general adaptability because of the methods we selected and the combination strategy we proposed and the use of metabolic treatment in the model. We contrast the new model with models we choose at the end of the paper. And then make a clear analysis and compare on the models. The conclusion shows that the new model is applicable for the real condition and has a better result.

In this paper, the slef-certified proxy signature scheme, which is proposed by Youan Xiao independently, is analyzed and the scheme turns out to be insecure. Concretely, there are two flaws in their scheme: Firstly, in the digital signing power delegation phase, the proxy secret key is generated without the use of the secret key of the proxy signer. Then, the proxy signer can disavow the signature which is really done by him, because the proxy secret key and proxy public key can be generated by the original signer only. Furthermore, we also point out that such flaw also exist in some other papers. Secondly, the self-certified proxy signature scheme can suffer from the public key replacement attack so that any one can forge a valid signature on any message. Finally, to overcome the above two flaws, an improved scheme is proposed.

Considering the features of cloud computing, a new web service migration infrastructure is presented in this paper. The proposed infrastructure takes some essential factors into account, such as server loading, service type, and so on. The work in this paper includes two parts: (1) We introduce an infrastructure for web service migration in cloud. Due to the properties of cloud, we modify and expand the SOA framework. (2) This paper presents an efficient method for establishing a schema of migration decision. The schema is designed to be extensive enough for different providers to determine the tradeoffs on service selection and migration. The application of this infrastructure can minimize the communication cost and improve the service performance when the service providers are widely distributed in clouds.

This paper discusses the problems appearing in the current renewal of urban historic site by computer-aided design from the cross perspectives of environmental design, architecture and computer science. Based on the discussion of the relationship between the research of renewal of urban historic site and computer-aided design, the paper focuses on how to use computer-aided design during two important stages which are noumenal form of historic site and comparison in the process of design and attempts to explore a systematic and scientific research method. The case studies on renewal conceptual design in Taosha Alley in Wuxi by the use of computer-aided design. Through the analysis of the case,this paper aims to provide a convenient, intuitive, complete, systematic design thinking and reference for similar studies.

In this paper, we give an overview of methadone treatment in Ireland and outline the rationale for designing an electronic healthcare record (EHR) with extensibility, interoperability and decision support functionality. Incorporating the open EHR standard, we designed a set of archetypes in line with the current best practice and clinical guidelines. They guide the information gathering process. A web-based data entry system has been implemented, incorporating elements of the paper-based prescription form, while at the same time facilitating the decision support function. The data collected via this Irish system can be aggregated into a larger dataset, if necessary, for analysis and evidence gathering, since we adopted the open EHR standard. It will be later extended to include the functionalities of prescribing drugs other than methadone during our work at the HRB Centre for Primary Care Research [1].

Since the network architecture is confronted with some serious challenges on security and management, it is necessary to develop a new architecture with credibility and management. However, most of the current studies redesign the existing network architecture at great expense. In this paper, the authors propose a new cognitive and credible network architecture that has the feature of enhanced cognitive management node. The original network function module can be reclassified, the user behavior can be expected and be managed, and the cognitive support can be provided under the new architecture. Also, the architecture is discussed in cognitive ability, credibility capacity, and mobility. Finally, further work is expected.

The recognition of prosodic structure is an important research aspect in the field of Text-to-Speech. It is essential to improving the naturalness of machine-synthesized speech. This paper proposes an approach to predicting and assigning prosodic structure automatically for Chinese sentences based on their tree structures. It presents the modeling of a statistical language model based on the simply handled corpus, the generation arithmetic of trees, and the method for predicting prosodic boundaries based on tree structures. The experimental results show that this tree-based method is effective, and it can obtain good prediction results.

The prepaid service is a valid method to assure the credit of users that is prevalent used by vendors at present, and it is a very important and applied function. Firstly, the generic AAA architecture and the work principle of RADIUS protocol used by prepaid service are introduced simply. Secondly, the system structure of VoIP authentication and accounting system based on RADIUS is described. Finally, this paper emphasizes on the implementation of the VoIP prepaid service. The results show that the prepaid service can enlarge the function of VoIP authentication and accounting system and validly control the customer credit.

Aiming at the shortages of traditional intrusion method, this paper presents a new kind of method for network intrusion detection-AMID (Intrusion Detection Based on Average Membership), and its algorithm. By measuring the close degree between the real-time behavior and normal behavior, the proposed method judges if there are intrusions. The experimental results show that the proposed method has considerable improvement in decreasing the false positive rate.

This paper presents a practical solution for secure communication by controlling a multi-chaos system via key-drive technology, and by combining classical encryption algorithm and dual chaos encryption algorithm with additional robust-gain and load-balance control. Signal is first encrypted using chaos random iterative encryption. Then it is further encrypted using chaos key control system. Meanwhile, to protect against attacking based on prediction method, multiple non-linear variation is used to encrypt the modulating signal, which will add new cipher code. For receiver, received signal is decrypted using inverse transition to the sender. Then the decrypted signal is sent to a responsive system designed using discrete system synchronization method based on PC synchronization and feedback synchronization. Finally, the original signal is recovered using chaos iterative decipherer by extracting iteration times and initial value from the decrypted signal.

Peer-to-Peer (P2P) media streaming networks have attracted significant research interest recently. However, it is challenging to design live P2P media streaming systems because of the stringent time constraints on the delivered media streams, which require more efficient and resilient overlay. In this paper, we focus on live P2P media streaming networks, a promising application flourishing in the Internet and which requires the distribution of live multimedia content to subscribers. We discuss several live P2P streaming systems, present a basic taxonomy of P2P overlays and closely examine these three approaches, tree-based, mesh-based, tree-mesh-hybrid-based, and summarize the major issues about them. we also categorizes peer selection schemes which are essential for the construction of a P2P streaming network.

A new logical method based on game to model and analyze electronic payment protocols is proposed in this paper. Strict formal analysis for Bolignano protocol is made by this new method, and Bolignano protocol is discovered non-fairness. These works indicate that the ATL logic based on game is more suitable to describe and analyze electronic payment protocols than traditional CTL.

The application of clipping general polygons in regular grids occurs frequently in geographical information systems (GIS). In general terms, the polygons usually clip into rectangle one by one in the pretreatment section, which need high-performance computer and long-time running. And also in this condition the pretreatment section needs low-CPU and low-memory usage. To improving clipping general polygon in regular girds, a new algorithm is presented in this paper. We use a new successive encoding base on Cohen-Sutherland line clipping algorithm to clip the general polygon in regular grids. We call it Grid-Clipping algorithm. The algorithm is simple and easy to understand and implement. Lent n be the total number of edges of all the result polygons in the girds, our algorithm computes the operation in time O(n).

Software is expected to be derived from requirements whose properties have been established perfectly. However, requirements are often inaccurate, incomplete or inconsistent as it is a very difficult task to define and analyze requirements. On the other hand, programs most likely deviates from requirements during implementation as the result of misunderstanding or/and neglecting requirements of software engineers. Deviations between programs and requirements are error prone, or cause software to act in unpredictable or unexpected ways. In this paper, we propose a novel framework that uses graph-based mining techniques to discover software execution patterns from object graph firstly, and then searches and matches within a pattern repository to determine whether the discovered software execution patterns are potential deviations from requirements corresponding to neglected requirements or not. After that, the new discovered software execution patterns are labeled and saved back into pattern repository. Hence, the framework is evolutionary and its ability will be more powerful. We give a case study to show how the framework works. The work indicates that the framework is effective and reasonably efficient for improving software quality.

This paper describes the reliability and validation of prediction models of LAN/WLAN integration network. An improved PSO algorithm is used to optimize the weight of BP neural network. Support vector machine (SVM) is used in network reliability prediction. The LAN/WLAN integration network reliability prediction models are established with three methods (BP neural network, improved BP neural network based on PSO, and SVM regression model). The validity of reliability prediction model based on artificial intelligence as well as the advantage of using support vector regression method has also been demonstrated experimentally.

In order to realize fast location and detection of abnormal status during running of industrial furnace, especially abnormal status of firing, this article studies and designs a fault diagnosis expert system based on fault tree theory. Firstly, formalized definition of industrial furnace fault diagnosis expert system is given in the paper, then all component elements of the expert system are analyzed and designed in detail, finally the principles and methods of design knowledge base are introduced to us by using the fault tree theory, and also reasoning algorithm is put forward, which is used to reason the fault by way of fault tree. The project practice indicates that this system knowledge model has good suitability, and what's more it is used simply and conveniently, also the outcome of fault diagnosis is reliable and steady.

Relevance feedback is an iterative search technique to bridge the semantic gap between the high level user intention and low level data representation. This technique interactively determines a user's desired output or query concept by asking the user whether certain proposed 3D models are relevant or not. For a relevance feedback algorithm to be effective, it must grasp a user's query concept accurately and quickly. In this paper, we propose a relevance feedback framework based on Bayesian logistic regression in content-based 3D model retrieval systems to incorporate relevance feedback information. Bayesian logistic regression relevance feedback framework using an active learning algorithm based on variance reduction to actively select documents for user evaluation. Experimental results show that this algorithm achieves higher search accuracy than traditional query refinement schemes.

A method that can fast, real-time get the projection matrix in ART reconstruction is proposed based on helical cone-beam CT trajectory in this paper. For a given ray this method can rapidly acquire the indices and corresponding lengths of the voxels passed through by this ray through sorting the intersections between the ray and mesh planes. Meanwhile we only store the projection matrix elements required for a given ray and release them after using them. By doing so the computation time and the storage space are significantly reduced. The experimental results show that this method can heighten the speed and better suppress noise.