Applied Mechanics and Materials Vols. 631-632

Abstract: Matrix factorization is a collaborative filtering recommendation technique proposed in recent years.In the process of recommendation,each prediction depends on the collaboration of the whole known rating set and the feature matrices need huge storage. So the recommendation with only one node will meet the bottleneck of time and resource. A MapReduce-based matrix factorization recommendation algorithm was proposed to solve this problem.The big feature matrices were shared by Hadoop distributed cache and MapFile techniques.The MapReduce algorithm could also handle multi-λsituation.The experiment on Netflix data set shows that the MapReduce-based algorithm has high speedup and improves the efficiency of collaborative filtering.

Abstract: Based on Entropy and dissipative structure theory, the structural model of social information field space is explored in this paper. Studies have shown that a relatively independent complex social system is posed by all kinds of "information men". And because of interconnected relationship and mutual transformation among "information men", it develops into a social group or social organization with the characteristics of mathematics.A complex system of society is finally regard as a "superior" constituted by different social groups according to certain rules.

Abstract: This paper introduces the complexity of fault information and summarizes some main methods in the field of fault diagnosis,including Expert system,Artificial neural network,Fuzzy inference,Rough set theory,Petri network,Bias network,etc.Analyse basic concepts,method principles and their advantages and disadvantages of various methods.Put forward methods of combining multiple methods to synthetically handle the fault information of power network

Abstract: The present research work proposes a method of optimal control for a class of nonlinear singular system, which applies single-step design, Zubov’s method and optimal control theory. Firstly, it makes the nonlinear singular system normalizable, realizes feedback linearization and the pole placement in a single step. Then, it constructs a performance index that is calculated explicitly as an algebraic function of the controller parameters by solving Zubov’s partial differential equation. Lastly, standard optimization techniques are employed for the calculation of the optimal values of the adjustable parameters. So we obtain the single-step optimal design of normalizability , feedback linearization with the pole placement for nonlinear singular system. The proposed approach is finally applied in a SIS model with logistic growth. The simulation result shows the feasibility and the effectiveness of the proposed method.

Abstract: Knowledge community, an online community form, is different from the real ones. It emphasizes the importance of effective knowledge management and the trust among the members. Researches on trust nowadays mainly focus on the P2P business model, the personalized recommendation systems of mobile equipment and etc. This paper sought to find a way to quantify the trust of knowledge communities. Two cases are considered in the paper. In this paper, it is thought that trust can be calculated based on the historical interactions, and if not, it can be predicted by user’s profession and preference. This paper provided two models based on the two different situations, and verified the model by an experiment.

Abstract: With the computer technology boomingly developing and the internet applications gradually spreading all over the world, more and more requirements of endpoint user acquires all kinds of resources via different security level networks. TCG provides approaches to prevent endpoint device from being ruined. However, more NARs on a single physical platform handle connections to different networks which lead to security problems, due to tightly coupled of NARs. In this paper, a lightweight trusted connect framework based hardware-level virtualization is proposed. And we illustrate core implement with improved KVM.

Abstract: Combined with the Map/Reduce programming model, the Hadoop distributed file system, Lucene inverted file indexing technology and ICTCLAS Chinese word segmentation technology, we designed and implemented a distributed search engine system based on Hadoop. By testing of the system in the four-node Hadoop cluster environment, experimental results show that Hadoop platform can be used in search engines to improve system performance, reliability and scalability.

Abstract: In order to process the abundant information in fuzzy clustering, one fault diagnosis method was proposed based on Rough Set reduction algorithm and Fuzzy equal relationship clustering. Not only the iteration numbers was reduced in the fuzzy equal relationship matrix, but also the clustering numbers was lower. Then the examples were applied to test its validity.

Abstract: This paper makes a brief introduction on the development and the major structure of Google glass, and discusses practical applications and a development trend of Google glass in the field of journalism based on the news business needs.

Abstract: With the development of Internet of Things technology, 3G communication technology,control technology and Multi-information fusion technology of multi-sensors, a technical solution for the robot to perform a task in a variety of complex, dynamic, uncertain and unknown environment is presented,which is based on these technologies and control theory, signal processing, artificial intelligence, probability and statistics. From the perspective of smart home,the robot technology and embedded technology, this paper presents an intelligent robot system which can interact with us through 3G mobile network.