Multi-attribute Decision Making For QoS Of Large-scale Web Service Composition Based On Ant Colony Algorithm

With the development and progress of software development and information technology,Web service technology is widely used in various fields.However,a single web atomic service is only developed to complete particular functions,which is difficult to adapt to more intricate user needs.Thus multiple independent services with limited functions can be combined to produce new composite services.Therefore,research problem arises: Web service composition based on Qo S,that is,how to select the appropriate service for each sub services of the composite service.With the rapid increase of the number of sub services and candidate services,it will develop into a large-scale web service composition problem based on Qo S.In addition,how to choose the best candidate service combination that best meets the needs of users is also a key problem.In this thesis,these two problems are transformed into multi-attribute decision-making problems.The multi-attribute and ant colony algorithm are combined to solve the problem of large-scale Web service composition based on Qo S.Through the use of various forms of evaluation information and the ideal point method,the evaluation decision of candidate service composition is accomplished,which makes the decision result more scientific and dependable.The following works are completed in this thesis:Firstly,the correlative research of web service composition is recommended as well as the ant colony algorithm and its advantages and disadvantages.Also,a few expression methods of evaluation information and their normalization methods of are discussed as well as the theoretical basis of the calculation method of objective attribute weights.Next,this thesis recommended the research on using ant colony algorithm to solve the problem of large-scale Web service composition based on Qo S,and the devise of state transition probability and pheromone renew formula in ant colony algorithm is proposed.An example is used to verify the practicability,research value and meaning of ant colony algorithm in solving the problem of large-scale Web service composition based on Qo S,and the simulation is carried out by MATLAB.Then,the selection criteria of Web service composition evaluation indicators are studied,and the interdependent evaluation system is constructed.The evaluation information of different expression forms is normalized by deviation transformation method,proportion transformation method and center of gravity method,which are transformed into single point value form to acquire a normalized decision matrix.Finally,The objective weight of attributes is acquired by entropy method.Multiple decision makers express their opinions on attributes in different forms,which are normalized and aggregated by using geometric average method to obtain the subjective weight of attributes.The subjective and objective weights are weighted and integrated into the comprehensive weight of attributes.The ideal point method is used to acquire the overall values of candidate service combinations,so as to sort them.An example is given to prove the practicability and dependability of the algorithm.The problem of large-scale Web service composition optimization is solved in this thesis by means of the ant colony algorithm.Furthermore,based on evaluating and ranking the candidate service compositions,a service composition can be obtained to better meets the needs of users.