Research On Composition Web Service Decision Makeing Approach For Complex Context

Nowadays, an increasing amount of companies and organizations implement their core business and outsource other applications as Web Services over Internet. Single Web service cannot satisfy the user's advandced demands because its function is sample. The composition Web service can provide complex functions, so how to create robust service compositions becomes the next step work in web services and has attracted a lot of researches. At the same time, composition Web services selection based on context plays an important role in web services composition.Our work focuses on the Web service composition for complex context. Within our knowledges, researches on this field are underway. Several scientific problems remain:1. There isn't a good context model to describe the context related on Web service composition and excellent service composition framework for context.2. The composition Web service selection algorithms can not cope with hybrid context matching, the QoS calculation with time dimension, and the composition service selection with uncertain QoS values (real numbers, interval numbers, triangular fuzzy numbers and intuitionistic fuzzy sets).3. Run-time performances of existing composition Web service seleciton algorithms decrease severely when tasks and candidate services increasing.4. Can not support the composition Web service selection for large group.5. Can not support the composition Web service selection with multiple decision-makers and heterogeneous QoS models. For solving the difficulties above, this thesis introduces ontology model for complex context and context-based framework for composition service selection, a hybrid context-aware dynamic composition service selection algorithm, a composition Web service selection strategy for group Pareto optimal plan and a compositon Web service selection algorithm based on multiple decision-makers and heterogeneous QoS model. Our research work has several significant contributions:1) Constructing the complex context ontology and composition Web service selection framework. In ubiquitous and converged network environment, the composition Web service selection must be considered the user's context and the running conditions for composition Web service, and the context's data types include real number, interval number, and language phrase. Especially in the ubiquitous, and converged network environment, group activities can be seen everywhere. Therefore, new context ontology for Web service to be established is an urgent task. For the first time, this paper presents a novel complex context ontoly (OWL_S_QoS+) to solve the difficulties above. It includes user context ontology and Web service context ontology. Based on hybrid QoS model (real number, interval number, triangular fuzzy number and intuitionistic fuzzy sets), we propose a new composition Web service QoS aggregation model. The other contribution is this paper designs a complex context-aware composition service selection framework for the composition Web service selection problem with complex context.2) Presenting a hybrid context-aware dynamic composition Web service selection algorithm. In ubiquitous environment, Web service selection is still challenging because of network and resources change frequently, the capabilities of devices and terminals are quite different, and the data type of context information is hybrid. Multi-period QoS evaluations have to be considered to obtain a reliable decision in the service selection process. Run-time performances of existing composition service selection algorithms decrease severely when tasks and candidate services increasing. For the first time, this paper presents hybrid-context-aware dynamic Web service selection algorithm to solve the difficulties above. It includes five main steps:Hybrid Context Matching Algorithm (HCMA), Hybrid QoS Aggregation Operators (HQAO), Atomic Services Cluster (ASC), generate composite service plan, and Service Selection Algorithm (SSA). A set of experiments demonstrates the benefits and effectiveness of our approach.3) Presenting a composition Web service selection strategy for group Pareto optimal plan. Group-oriented services and composition services are widely applied in daily life. Group-oriented composition service selection is still challenging for big group and personalized QoS requirements. Therefore, this paper presents a large Group Pareto optimal Service Selection Algorithm (LGP_SSA) based on multi-attribute group decision making (MAGDM) theory. LGP_SSA includes five main steps:calculating group preference, calculating weighted cluster weights, constructing cluster weighted normalized decision-making matrix, determining the group positive-ideal and group negative-ideal solution, evaluating alternatives for group synthetically. Experimental results show that the proposed algorithm can solve the problem of composition Web service selection with big group and personalized QoS requirements very well.4) Presenting a composition Web service selection algorithm based on multiple registration centers (multiple decision-makers) and heterogeneous QoS model. Based on Multi-Attribute Group Decision-Making (MAGDM) theory, this paper presents a novel Multiple Registration centers and Heterogeneous QoS model's Service Selection Algorithm (MRHQ_SSA) to solve the above difficulties for the first time. MRHQ_SSA not only can support the composition Web service selection with multiple decision-makers and homogeneous QoS model, but also can support the composition Web service selection with multiple decision-makers and heterogeneous QoS model. Through evaluating the QoS information of all QoS registration centers, MRHQ_WSSA can obtain the synthetically optimal composition service plan. It includes six main steps:normalizing QoS model, normalizing the user's QoS preference, calculating the group preference, weighting normalized decision-matrix, determining group ideal solution and calculating the relative closeness to the group ideal solution. And a set of experiments demonstrate the benefits and effectiveness of the approach.