Earticle

Trust in e-commerce has become one of the most important issues in online applications. Constantly, a user will only search for the most credible of goods and service providers and then take on their transactions. How to confirm which service providers are the most trusted for a user has become the most critical problems. This paper presents a trust network and small world trust community clustering for the analysis of the users most trusted relationship. It uses the nodes to represent the various subjects involved in the trust and use the connection links to denote relationships. The weight of the links indicates the strength of the relationships. First, it construct a trust network diagram which has the weight value of links, and then to analyze the clustering properties of the relationship according to the weights and the path length. At last, it classifies the most trusted subjects to the same cluster for a user. Local trust recommendation degree and global trust recommendation degree are used to evaluate trust relations among subjects and it gives an improved shortest path algorithm to construct trust network. A clustering algorithm based on coefficient and path length is presented for e-commerce trust network community. Experiments show that the method of building trust through the network model can well describe the main indirect e-commerce trust and the algorithm has obvious advantages in accuracy and time cost.

Files are created for Traffic Analysis, Maintenance, Software debugging, customer management at multiple places like System Services, User Monitoring Applications, Network servers, database management systems which must be kept for long periods of time. These Files may grow to huge sizes in this complex systems and environments. For storage and convenience files must be compressed. Most of the existing algorithms do not take temporal redundancy specific Files into consideration. We propose a Linear Cellular Automaton based Classifier which introduces a multidimensional File compression scheme described in eight variants, differing in complexity and attained compression ratios. This scheme introduces a transformation for File whose compressible output is far better than general purpose algorithms. This proposed method was found lossless and fully automatic. It does not impose any constraint on the size of File.

The trust-aware recommender system (TARS) suggests the worthwhile information to the users on the basis of trust. Existing models of TARS use personalized rating prediction mechanisms, which can provide personalized services to each user, but they are computational very expensive. We therefore propose an efficient global rating prediction mechanism for TARS: we use the k-shell decomposition to find the most influential nodes in the trust network, and use the recommendations given by these nodes to predict global ratings on items. The experimental results verify that our proposed method can predict ratings accurately with low computational complexity.

In order to maximum the hardware resource utilization, Virtual Machine (VM) management has become an important research field of virtualization technology. VM scheduling is crucial for the throughput of a system and thus affects the overall system performance. According to the scheduling algorithm of Credit Scheduler using in XEN hypervisor, each virtual CPU is asynchronously assigned to a physical CPU by CPU scheduler in order to maximize the throughput. But for a concurrent program, the implementations of threads are not completely independent. In this paper, a solution is proposed for the synchronization problem of a concurrent program. By modifying the Xen Credit Scheduler, threads in the concurrent program can be synchronized and the performance in concurrent workload is greatly enhanced. The waste of CPU time can be reduced and achieve higher throughput while keeping the design of the scheduler to be light weight, fair, and efficient. With the proposed scheduler, more VMs can be deployed while achieving the same throughput, thus gain greater utilization of resources and better energy efficiency.

Bankruptcy prediction has been addressed by many researchers in the field of finance since few decades. One of the best approaches to deal with this issue is considering it as a classification problem. In this paper a time series prediction model of bankruptcy via Adaptive neuro-fuzzy inference system (ANFIS) is formulated, which is capable of predicting the bankruptcy of a firm for any future time. The data used in this study has been extracted from the past financial records of ongoing and failed enterprises. The extracted financial ratios are preprocessed by calculating Altman’s Z-score before feeding into Time series model. The Time series prediction is carried out using ANFIS to predict the bankruptcy at any given time which overcomes the limitation of Altman’s basic model of bankruptcy prediction. Fuzzy Logic Tool box of MATLAB has been exploited for simulations and evaluation of the model. Numerical illustration is provided to demonstrate the efficiency of proposed model.

Cyber physical systems(CPS) include a lot of high complexity computing such as physical aspect modeling, dynamic analysis and verification of continuous dynamic property, analysis and verification of real-time property, analysis and verification of spatial property, scheduling and non-functional requirements. The correctness of computing results of cyber physical systems not only meets the time requirements, but also meets the spatial requirement, this make that it is impossible to solve the several difficult computing problems by traditional methods. In this paper, we describe some of the research directions that we are taking toward addressing some of the challenges involved in building cyber physical systems. The ultimate goal is to develop foundations and techniques for building safe and effective cyber physical systems.

Smart card based password for authentication has become a common trend. Although smart card brings conveniences, it also increases the risk in the case of lost cards. In other words, when the smart card is possessed by an attacker, the attacker will possibly attempt to analyze the secret information within the smart card to deduce the authentication mechanism of the server and then forge user credentials or break the entire authentication system. In this paper, we analyze the lost smart card attack from Juang, et al.’s scheme [9] that proposes password authenticated key agreement and propose an improved robust and efficient user authentication and key agreement scheme using smart cards. In order to bolster the security of the entire system, we mitigated some of its weaknesses.

Traditionally it’s impossible to monitor the freight’s status inside a container which is made of steel and makes data transmission by radio impossible, let alone the wired communication solutions. As container transportation is the prevailing method for international cargos, a feasible way to transmit status of the monitored cargo out of the stacked containers is very important, especially for ones with special requirements such as dangerous goods. This paper proposes a scheme to construct a communication network for stacked containers by deploying communicating nodes on the container’s surface to form a customized wireless sensor network. An effective routing algorithm is also proposed to exploit the characteristic of the network which is composed of smart containers. Simulation results show that the proposed interconnecting scheme makes an effective communicating infrastructure for upper layer container management services.

Loan default evaluation and discrimination is a complicated issue because of its nonlinearity and uncertainty. Least square support vector machine (LS-SVM) has been successfully employed to solve regression and time series problem. This paper proposes a novel PSO-LS-SVM model based on the improved PSO algorithm to optimize parameters of LS-SVM, which is a new improved form by synthesizing the exiting model of PSO. Some evaluation indices, which are reduced without information loss by a genetic algorithm, are used to train PSO-LS-SVM and discriminate between healthy and default testing samples. A case study based on financial data acquired from listed companies has been carried out. Result has shown that the proposed model has a distinct improvement in the aspect of accuracy rate as compared to PSO-SVM, LS-SVM, SVM and BP neural network.

Alternative splicing (AS) and Gene duplication are two well known evolutionary and ubiquitous mechanisms. Both have the common gist for conveying the functional diversity by means of escalating gene variegation. The objective of this study is to understand the established hidden relationship between both of these mechanisms to find the answer for the long standing puzzle circumventing the evolution of genome size while utilizing the computational power. We investigated that the alternative splicing can promise to unveil the nature as well as operation of cellular codes meticulously which play inevitable role of combinations of regulatory elements in pre-mRNA. This study also describe in detail of the cellular complements of splicing regulators, which collectively establish regulated splicing pathways. We have proposed a computational framework based on subset theory to address a part of the inexplicable functional diversity paradox which is also known as c-value enigma. The operational detail of this framework has been discussed to give an insight into the possible solution towards the problem of more proteins than their corresponding genetic material.

As an important decision support query, Group Nearest Neighbor (GNN) query has received considerable attention from Location Based Service (LBS) research community. Previous works paid much attention to the uncertain data objects (P). Nevertheless, very little work has done to the scenario when query objects (Q) are also uncertain. In this paper, The Range-based Probabilistic Group Nearest Neighbor (in short RP-GNN) query is introduced to draw a comprehensive discussion for this extended scenario. Two novel pruning methods are proposed to improve the performance of RP-GNN. The effectiveness, efficiency and scalability of proposed methods are validated through extensive experiments. The proposed methods achieve an average speed-up of 62.2% against existing probabilistic GNN algorithms and 1-2 orders of magnitude against linear scan.