Showing papers presented at "International Conference on Service Sciences in 2014"

Artificial Bee Colony Optimized Scheduling Framework Based on Resource Service Availability in Cloud Manufacturing

Abstract: The research on Cloud manufacturing (CMfg) is mainly articulated around the promotion of collaboration among service providers to increase the global manufacturing capabilities and create virtual enterprises that satisfy complex service requirements and designs. For manufacturers, here denoted as service providers, CMfg also presents a valuable enhancement of their resources' occupancy and a way to rapidly expend their business. However, the interplay among service providers is an important parameter to issue when it comes to cloud service scheduling. The collaboration orientation and the resource occupancy must be addressed as the main driver for the scheduling framework establishment. Therefore, this paper proposes a strategy concerning the scheduling approach in CMfg based on resource service availability and globally optimized through Artificial Bee Colony (ABC) algorithm. The precision and efficiency of the present method are discussed in the experiments section.

A Survey and Taxonomy of Cloud Migration

Abstract: Cloud computing has certainly gained attention and skyrocketed in the technical and economic world because of the appealing features. With decades of development there are lots of on-premises applications and systems in use. Consequently, the demand of migrating on-premises applications and systems to the cloud computing is gigantic. Thus, the cloud migration is not systematically reviewed with a proper taxonomy due to the variety of cloud computing architecture and the complexity of applications and systems. This paper surveys the cloud computing architectures and cloud migration decision frameworks by both the industry and the academia. Then, it proposes cloud migration taxonomy for a clear understanding of related approaches. Finally, it addresses the future challenges and direction as well.

Enhancing Book Recommendation with Side Information

Abstract: Recommendation systems are being broadly adopted in various applications to suggest items of interest to users amidst the enormous volume of available information. And many academic libraries have implemented various recommendation technologies to attract more readers and evaluate the resource utilization. And collaborative filtering (CF) technologies are widely used. However, one key issue limiting the success of collaborative filtering in certain application domains is the cold-start problem. In this paper, we aim to solve this problem with side information including the profile of the readers and the information of the books. We propose three approaches: the first is a recommendation method based on readers' side information, the second one is based on the books' side information, the third one contains two methods to combine the side information and the rating information. And the experiments evaluated on the real dataset which is very sparse validate the efficiency of the methods.

Resources Allocation Method on Cloud Computing

Abstract: To improve the efficiency of resource allocation cloud computing, as while as to improve resource utilization for the service provider, the paper has raised a polymorphic algorithm of Ant Colony Optimization, which can assure the quality of cloud service, and also dynamically change the list that contain nodes information. When the user submits the task, the algorithm will transfer it to the Cloud Control Queen by Master. And then, according to the functions, the ant colony will be divided into test ant colony, reconnaissance ant colony, cleared ant colony and workers ant colony. The algorithm can achieve the minimum average completion time gradually, and may reduce local optima, by forecasting the completion time and other pheromone.

Context-Aware Trust Aided Recommendation via Ontology and Gaussian Mixture Model in Big Data Environment

Abstract: With the development of big data, the data size becomes bigger and bigger, which makes users consume enormous time to find the items that they might like from abundant options. Recommender systems are expected to help users find interested items. However, most existing recommendation methods do not take into account any additional contextual information with a reasonable complexity. This paper aims to propose a context-aware recommender system by incorporating context-aware technology into recommendation. The context-aware approach is based on ontology and Gaussian Mixture Model. The recommendation analysis is implemented by trust aided probabilistic matrix factorization approach. The evaluation shows that the proposed approach has a good effect in recommendation quality.

A Linked Data Based Personal Service Data Collection and Semantics Unification Method

Abstract: The flourish of services on the Internet results in the flourish of personal service data. The volume of the data produced by various services around a person is fairly large. However, in the service-centric paradigm, these data are distributed in the clouds of different service providers and these clouds are mutually isolated, but there lacks of a global view of the service data related to a specific person, limiting the collaboration between various services. To overcome this deficiency, we propose a user-centric paradigm for personal data management, called "Personal Data Cloud" (PDC), which is a SaaS application storing a user's data from different services in a centralized way. A key issue in PDC is how to collect personal data from distributed services and store them in a unified form. In this paper we present three ways of collecting personal data, namely open API based, web crawler based, and manual importation based ways. To deal with the semantics unification among the collected data, we adopt Linked Data and ontology techniques for semantic annotation. Some open APIs are designed for the PDC to facilitate the access and usage of the personal data from a portal or apps on mobile terminals.

High Availability Benchmarking for Cloud Management Infrastructure

Abstract: Cloud-management infrastructure plays an important role as a part of cloud computing stacks, serving as the resource manager of cloud platforms. The complexity of cloud-management infrastructure makes its high availability (HA) one of the most critical requirements. Various technologies have been developed to increase the reliability and availability of cloud management infrastructure, however, little work focused on quantitative analysis of its availability. In this paper, we designed a new availability benchmarking method for cloud management infrastructure. By using our measurement method, a private cloud customer or public cloud provider can estimate the availability of deployment of its cloud management stack. We have evaluated our method on Open Stack cloud infrastructure. We show that with various HA technologies and configurations, the availability of the cloud management infrastructure can be greatly different.

Privacy-Aware Web Services Selection and Composition

Abstract: Privacy protection is still a key challenge in the web service composition area. The user and web service provider frequently share their privacy data with other web service will increase the risk of misuse and disclosure of privacy. In this paper, we first analyze the privacy protection problem according to define four types shared data within web service composition, and then present web service composition and privacy models, our models allow user and web service provider to define user's privacy preference and web service's privacy policy specification. An algorithm is presented to check the policy compliance on the dependency edge and select a set of policy compliant web services with compositing a well privacy protection web service composition. Experiments demonstrate the applicability of our models and algorithm.

A Collaborative Filtering Recommendation Algorithm Based on User Preferences on Service Properties

Abstract: In the service recommendation, the data of user ratings are usually very sparse. In the case of data sparsity, item similarity which is based on user ratings in the traditional item-based collaborative filtering algorithm ignores the situation that users are different in regarding various item properties. That results in the low accuracy when predicting. Based on this point, this paper proposed a collaborative filtering algorithm based on user preferences on service properties to solve the data sparsity problem in the service recommendation. This method firstly builds the service property preference model for each user based on information theory. Secondly, computes the service similarity correction factors of each user on any two services with service properties similarity. And finally the similarity of two services is the sum of service similarity correction factor and the Pearson correlation coefficient of them. The experiment results suggest that the proposed algorithm can efficiently improve the recommendation accuracy in the case of data sparsity.

Collaborative Filtering of Web Service Based on MapReduce

Abstract: As more and more web services appear on the internet, it becomes more difficult for us to pick out a suitable service among a large number of alternative services. The services recommended by user-based collaborative filtering lack relevance, and it is insufficient to recommend the new services. In this paper, we proposed a collaborative filtering method mixed user-based and item-based collaborative filtering. In order to adapt to the era of big data, it was implemented making use of MapReduce framework. We avoid overestimated similarity and the sparseness to improve the algorithm. Experiment results show that the hybrid collaborative filtering method can not only ensure accuracy, but also provide chance to recommend the new services.

Petri Net-Based Context-Aware Service System Modelling: An Overview

Abstract: Context-aware service system provides personalized context-aware services to a user actively, according to the specific context information of the user, such as the position, the personal profile and the historical records. To build a formal model to simulate and verify the system is a crucial and key issue in context-aware service systems. Several approaches were proposed to build context-aware service system model with petri net. However, besides the basic petri net model, there are several extended petri net models that enhance the expressive power of the basic petri net in different aspects. In this paper, we summarize how different petri net models are used in modelling context-aware service systems, and try to figure out the advantages of some of the petri net models, so as to make it easier in choosing a suitable petri net model when building context-aware service systems.

A VOP-based Value Analysis and Optimization Method for Service Selection and Composition

Abstract: Service selection and composition are important issues to realize IT-enabled service systems. Service stakeholders obtain different service values in the solutions of service selection and composition with the same function but different QoS. The service value achieved from service systems is a critical factor to evaluate and to select the suitable services for the solution of service selection and composition. A value analysis and optimization method for service selection and composition based on value-oriented priority (VOP) is presented in this paper. Firstly, the quantitative analysis method of VOP-based service elements is presented, which includes value contribution analysis, sensitivity analysis and trade-off point analysis. Then the value optimization model for minimal cost is given for solving the problems between expected service values and the constraints of service systems. In order to solve this service value optimization problem, a VOP-based artificial bee colony (VOPABC) algorithm is proposed. Finally, the validity of VOPABC algorithm is proved through simulation experiments.

A Location-Based Services Selection Model and Algorithm with QoS Constraints under Mobile Internet Environment

Abstract: With the rapid development of communication technology and mobile Internet, selecting from mass Location-based Services to meet the needs for different users becomes the research hot spot. First, it starts from QoS constraints as well as user preferences in this paper, then describes a mobile service selection framework. Furthermore, a User Preferences Database is built in modeling user preferences. Users' feedback is adopted to update User Preferences Database, which makes the combination between QoS constraints and user preferences more compact. Then, a LBS selection model and algorithm are proposed separately, which considers the QoS attributes and user preferences. Finally, it takes Online Learning Resources Platform based on mobile terminals as a sample, and designs a mobile course resources selection system on campus. The results show that the proposed algorithm is precision and effective in solving the problem of service selection to support Location-based Services personalization.

Product Recalls as Service Failure: Revisiting Consumer Reactions

Abstract: Due to the increasing complexity of products and closer scrutiny by manufacturers and policy makers as well as higher demands by consumers, product-harm crisis are expected to occur even more frequently, since goods can be viewed as distribution mechanisms for services, product recalls can be viewed as kind of service failure. The paper took this perspective to visit consumers' reactions to service failure, proposed a conceptual research model.

Fast Human Detection Using LDA via L1-Norm

Abstract: Fast detection is of vital importance in human detection sometimes. Considering the high dimensions of the features widely used in human detection, it will severely slow the detection speed. Therefore, in this paper, we try to find a way by using Linear Discriminant Analysis(LDA) via L1-norm regularization to solve this problem. It reduces the dimension of feature from 3780 to 150 before classification, and gets a more fast speed then SVM and LDA, while keeps a competitive accuracy.

Energy Efficient Mobile Cloud Computing and its Applications

Abstract: In this paper, we investigate offloading policy for energy efficient mobile cloud computing. To minimize the energy consumption on the mobile device and the cloud, we propose a general optimization framework based on the characteristic of applications. Particularly, for delay-sensitive applications, we formulate a delay-constrained optimization problem, in order to reduce the energy consumption on the mobile device while meeting the time constraint. For delay-tolerant applications, we formulate a stability-constrained optimization problem, in order to reduce the energy consumption in the cloud while satisfying the queue stability. Based on the optimization framework, we present the offloading policy of typical mobile applications that are computation intensive, including video transcoding, object recognition, image retrieval and virus scanning. The proposed offloading policy can reduce the energy consumption on mobile devices and the cloud, which provides guidelines for the design of green mobile cloud.

Customers' Perspectives on ISO 9001 QMS Auditors' Personality Traits: A Preliminary Investigation from Taiwan's Certificated Companies

Abstract: Under the trend of global trade, there becomes no boundary among countries. Following the tendency for achieving good quality, not only on the quality focus of products, Taiwan's companies paid much attention to quality system certification. Achieving the ISO 9001 quality management system certificate become the first priority for the quality level recognition of those companies. During the certification process of the ISO 9000 QMS, the auditors played an important role. Therefore, their personality traits may affect the result of certification, and customer's satisfaction level. In this research, by applying questionnaire analysis, we tried to review ISO 9001 QMS auditors' personality traits from customers' viewpoints. A questionnaire with 15 questions was used to collect the data, both from "importance expectation" and "perceived experience" sides, from Taiwan's ISO 9001 QMS certificated companies. There're 77 valid responded questionnaires in our research. From Cronbach's coefficient, the dimensions design of our questionnaire was verified and had both good reliability and validity. By applying paired-t test under 95% confidence interval for all 15 questions, only question number 8 (Under the pressure of work, auditors can effectively cope with) showed no significant difference. From which we can find the common existing gaps between "importance expectation" and "perceived experience" for audited companies. Also, a model for ISO 9000 QMS auditors' personality traits was proposed. All 4 factors with grouped questions were "interpersonal relationship and interaction" with question 5, 3, 4, 9, "self-discipline and growth" with question 11, 15, 12, 14, 13, "stick-to and focus" with question 6, 10, 7, 8, "integrity and objectivity" with question 1, 2. Results from analysis of those questionnaires can provided as a reference base for the certification organizations for their auditors' qualification definition and training purposes.

A Dynamic Weight-Based Dynamic Requests Scheduling Model in Multi-core Web Server

Abstract: In order to improve performance of handling user requests, most of web servers adopt multi-core processors. However, traditional requests scheduling algorithms, such as FCFS, don't consider the characteristics of multi-core processors and the distribution of the dynamic requests service time. Therefore, the scheduling algorithms couldn't fully exploit the performance of multi-core processors. To solve this problem, we proposed the new requests scheduling algorithm based on the dynamic weights of request queues. Simulation experiments have been done to evaluate the new algorithm. The experiment results show that the proposed algorithm could solve the load imbalance between cores in long period and avoid the problem of ping-pong effect in multi-core systems.

A Medical Diagnostics Service Chain Optimization Method Based on Stochastic Petri Net

Abstract: Medical diagnostics involves multiple levels of modern clinical diagnostic devices and the process of medical diagnostics confront different decision for data transitions among multiple layers. In this work we analyze the challenges of modern medical diagnostics, which is an important part of public healthcare, and formulate this complex process as service chain. Based on stochastic Petri net, a modeling and optimization method is proposed for the medical diagnostics service chain. The case study shows that the competition relationship in the medical diagnostic process can be resolve through adding dummy transitions into the Petri net, which reflects adding appropriate diagnostic nodes and data transmission in the medical diagnostic service chain in practice.

A Deployment Scheme of Virtual Machines for Campus Cloud Platform

Abstract: With the continuous deepening of teaching informatization, campus cloud platform is becoming increasingly popular, but the resource utilization in the campus cloud is still low, resulting in a serious waste of resources. To tackle this problem, this paper puts forward a deployment scheme of virtual machines for campus cloud platform, defines the course requirement model and the physical machine load model, and proposes a virtual machine deployment algorithm. This scheme can adapt to the characteristics of teaching applications such as periodicity, predictability, and batch, and achieve the purpose of energy saving and load balancing. Experiments show that the scheme can effectively reduce power consumption and achieve load balancing.

Applying MapReduce Framework to Peer-to-Peer Overlay Network

Abstract: MapReduce is a programming framework widely used in cloud computing environments for processing large amount of data in a highly parallel way. However, current MapReduce model do not cope well with its scalability, which means that under certain hardware configuration, it can only support limited scale of cluster due to the overloading of center node. In this paper, we present a prototype based on DHTs Peer-to-Peer MapReduce system, which removed the MapReduce task centralized scheduling's master node and bottom file system management's name node on the basis of remaining original MapReduce workflow unchanged. In the system, the distributed file system in bottom layer queries data through distributed hashing, while the MapReduce system in upper layer invoke and schedule the tasks by distributed notification mechanism. In this way, the system can theoretically achieve the scalability of Peer-to-Peer system. The scalability evaluation of the system has been experimented in the network scenarios using the prevailing word count problem.

Deviation Methods via Social Review Graph in Evaluation Systems of Online Store

Abstract: Evaluation system is an important issue in data mining. In evaluation system, all reviewers try to assign fair scores on a set of object. We want to obtain fair reviewers for all objects. However, it is the real fact that reviewer may deviate in their score assigned to the same object. Therefore, the deviation is unavoidable in evaluation systems. In this paper we define deviation of evaluation systems. Then, two novel methods are constructed to measure deviation: deviation from Correct-Core viewers and deviation from MaxGap-Coreviewers. The experimental results show that the proposed measure methods are efficient and can be used to solve deviation problem in evaluation systems effectively.

A Multi-properties Based Description Method of Cloud Manufacturing Services

Abstract: To improve completeness and availability of cloud manufacturing service in cloud manufacturing platform, a multi-properties based cloud manufacturing service description model was presented through analyzing its classification and characteristics. The model consists of five properties: characterization properties, functional properties, commercial properties, resource properties and technical properties, which was also described by using formal language. Based on the concept model, the cloud manufacturing service model was realized by Web Ontology Language for Service (OWL-S) and finally a prototype system was developed based on the cloud manufacturing service platform for small and medium enterprises, which verified the model's feasibility.

Service Selection Based on Dynamic Group Trust in Social Network

Abstract: People are more likely to have a common topic together, whereas gathering the people to form into the group also can create more value. Service network is the society that has different layers, where people are belongs to different kinds of interests and hobbies, therefore, group mode is very suitable for services network conditions. It is easy to form variant crowds of people and motivate common topics according to characteristics of population. Aiming the shortcoming at the aspects of social network of the existing service selection, this paper proposed a service selection algorithm and related group trust model in social network. Through calculation of group trust and its propagation between groups, it got the comprehensive trust evaluation. Through the correction of trust evaluation, it succeeded to resist some malicious behaviors. Finally, we verified its effectiveness, resist aggression through the simulated experiment.

An Optimal Service-Selection Model Based on Capability and Quality of Resource Service

Abstract: Most of the researches on optimal service selection are based on the assumption that the capabilities of the services fully meet the requirements. Their limitation is the ignorance of the resources which is the basic factor supporting the implementation of services and it may cause a waste of resources. In cloud computing environment which benefits from its large-scale, there are a large number of resources. Therefore, the waste of resources in it would be a big problem. This paper introduces 'service equivalent' as the basic metric to measure the capabilities of service resources and proposes an optimal service selection model based on capability and quality of service resources and algorithm, in order to solve the issues about the matching capability of service resource and the optimal selection of service resource based on quality. Finally it proves that the model can effectively reduce the waste of resources by the test, which achieves the expected goal.

Architecture Design for Service Network Deployment, Customization and Execution

Abstract: Service Network (SN) is a feasible approach to deal with mass personalized customer requirements in a cost-effective way. The SN philosophy has been widely applied in various service domains, such as e-Business, online travel services, bioinformatics, etc. Being a persistent service infrastructure, SN requires a run-time environment which enables its deployment, customization and execution. This paper shows an architecture design for such kind of environment. The conceptual model of SN is firstly introduced, and then the high-level system architecture is put forward, including a deployment engine (SDE), a customization engine (SCE), an execution engine (SEE), and a set of repositories storing necessary data during the lifecycle of an SN. Particularly, three mechanisms facilitating the execution of a customized SN are proposed, i.e., BPEL-based, Rule-based and Event-based ones, and a brief comparison between them is conducted. The proposed architecture enables the deployment and execution of SN across boundaries of different clouds where the constituent services of the SN are deployed and executed.

Seal Extraction Based on Local Thresholding Techniques and Color Analysis

Abstract: In this paper, we propose a method to extract seal imprints from bank check images. Because seals are often colorized, we can utilize color information to extract them. Firstly, we make use of local thresholding techniques to distinguish seal imprints from background areas. This step not only reduces the complexity of the extraction problem but also gets clear edges. Secondly, we utilize R and G components to get rid of machine-printed characters and noise. Finally, the results of our method are compared with other traditional methods. The experiments prove that our approach is capable of extracting Chinese seal imprints in most cases.

When are Referral Reward Programs Effective: A Perspective of Regulatory Fit Theory

Abstract: Many firms are now using referral reward programs (RRPs) to harness the power of word of mouth and to increase referrals to acquire new customers. However, a persistent problem is that rewards are not always effective. By applying Regulatory Fit theory, to understand how to sustain goal orientation (promotion or prevention) and stimulate word-of-mouth through a match with the message frame, this study demonstrates that the presentation style marketers adopt influences consumers' perceptions of reward value. One laboratory study demonstrates that fit effects result from the interaction between presentation format of reward (i.e., Gain vs. Loss) and regulatory focus and affect referral likelihood. In turn, the findings demonstrate that RRP efficacy can be enhanced by stimulating regulatory orientations that match the presentation formats of the reward.