Advanced Materials Research Vols. 756-759

Abstract: Euclidean distance measure is computationally simple and commonly used in the task of classification. However, it does not capitalize on any discriminant information from training samples, which is crucial to classification performance. In this paper, we learn a Discriminative Isotropic Space from both labeled and unlabeled data to improve classification accuracy as well as generalization ability. Our regularized objective function implicitly minimizes mutual information between input space and transformed space. There is no local optimum problem in our algorithm and its closed form solution can be easily obtained. Experiments on real data sets demonstrate the efficacy of our method.

Abstract: Based on studying Digital Urban Planning spatial database, this subject uses spatial data engine ArcSDE as interface between GIS application server and database server, and takes the ArcSDE as the core to realize spatial query and spatial analysis of digital urban planning spatial information, and then unified managed the spatial data and attribute data of digital urban planning, finally to support efficient, the huge amount of data extraction.

Abstract: With the rapid development and evolution of Internet, Web applications play a significant role in people's daily life and daily work. Usually, Web developers design the structure of the web application according to their experiences. But as the evolution of web applications, the existing structure is not enough to meet the needs of the users. This paper proposes an approach to reorganizing the structure of web applications dynamically based on Web logs. Obtained structure of the web application is more reasonable, more convenient to provide services for the users.

Abstract: In this paper, we propose a Bayesian network model. Firstly, the Bayesian network model is introduced, and Belief Propagation (BP) algorithm is utilized for model estimation. Then ExpectationMaximization (EM) algorithm is used for parameter estimation of the Bayesian network. Finally, the SAR image is segmented by calculating the Maximum Posteriori Probability (MAP) of each pixel. Experimental results show that, comparing with the Markov Random Field - Intersecting Cortical Model (MRF-ICM), our Bayesian network model gives better results in both segmentation and time-consuming.

Abstract: Facing the current inadequacy of informatization in walling management, this paper focuses on the design to use cloud computings reliability, secure data storage, convenient web services and strong capability of calculation. In addition, this paper sets up a model based on cloud-computing for walling management in governments. Firstly, the paper briefly introduces cloud computing, and then analyzes the advantage of walling management platform. Moreover, this paper points out the key issues should be solved. It will bring the positive effect of walling management informatization to government of China.

Abstract: This paper first discusses the related concepts of software component technology based on software component technology and software development method, and then based on the UML technology component extraction method and component extraction strategy, the system, the component extraction, the establishment of system function of each part and component model, software component technology is introduced into the online test system development, so as to improve the application software system development speed and efficiency and maintainability, reconfigurability and scalability.

Abstract: Quality of service (QoS) generally assumes more than one QoS measure which implies that routing can be categorized as an instance of routing subject to multiple constraints: such as cost, delay, bandwidth, etc. The problem of constructing multicast trees is studied to meet the QoS requirements where it is necessary to provide bounded constraints among the source and all destinations while keeping the cost of the multicast tree low. So, a kind of source-destination QoS multicast routing problem is addressed about communication networks. The algorithm we presented takes bandwidth, delay and loss rate as premise, constructs routing selected function based on shortest path, modifies selected path according to the function above so as to fit multi-QoS parameters. Simulation results show that the algorithm has both lower delay and better performance and can be extended to cases of multiple QoS parameters conveniently.

Abstract: Deep web data integration has become the center of many research efforts in the recent few years. Near duplicate detection is very important for deep web integration system, there are seldom researches focusing on integrating deep web Integration and near duplicate detection together. In this paper, we develop a integration system, DWI-ndfree to solve this problem. The wrapper of DWI-ndfree consists of four parts: the form filler, the navigator, the extractor and the near duplicate detector. To find near duplicate records, we propose efficient algorithm CheckNearDuplicate. DWI-ndfree can integrate deep web data with near duplicate free and has been used to execute several web extraction and integration tasks efficiently.

Abstract: Software development method based on component is currently a popular software production technology, its core around the component development and assembly technology. But how to combine the practical application requirements implementation based on component composition software development process is a worth studying further. According to the development of component technology, put forward a kind of network learning platform development framework based on component assembly, and study the website of component development and assembly method, apply this technology to the development of our online learning platform, results show the effectiveness of the proposed method.

Abstract: With growth of internet and social network service based on web, the complexity and the diversity of the cyberspace is increasingly focused by researchers. In order to research dynamic social interactions relations in cyberspace, this paper present a novel ideas to study cyberspace from new viewpoint of ecology. We first discuss the motivation from plant ecology to cyber ecology and redefine cyber ecology, which is different to previous definition, and then we present cyber ecology and computing research issues from individual, group (community), cyberspace environment and representing cyber information and knowledge. Finally, this paper compares cyber computing with existing research, such as social computing, urban computing and pervasive computing..