Informatics Publishing Limited

F. Ezhil Mary Arasi ¹, S. Govindarajan ², A. Subbarayan ²

Affiliations
1 Department of Computer Science and Engineering, SRM University, Kattankulathur, Chennai - 603203, Tamil Nadu, IN
2 Department of Master of Computer Applications, SRM University, Kattankulathur, Chennai - 603203, Tamil Nadu, IN

Abstract

Objectives: In recent times Web Service providers are finding difficult to satisfy the requirements of the consumers due to several reasons. We have made an attempt to identify the specific parameters for evaluation of an efficient functioning of Composite Web Services. The study is based on a data set sampled from a large number of Web Service Providers. Methods/Statistical Analysis: Random sampling technique was adopted for the preparation of the data set. A multivariate statistical technique viz., Factor analysis is used for identifying the factors. Factor analysis method reduces large number of variables into few factors without sacrificing much of explained variability by the variables. Factor analysis begins with the construction of new set of variables based on the relationship in the correlation matrix. Based on principal component analysis we have identified the factor which determines the effective performance of Web Service providers. Findings: The study is based on a data set sampled from a large number of Web Service providers. The results based on the sample data set indicates the components of Web Service Providers Performance factors, Security factors and Trust factors viz., Successability, Best practices and Throughput respectively. These parameters have the higher loadings compared to other extracted QoS parameters. Application/Improvement: The research work was confined to the usage of factor analysis method based on principal component analysis approach. As a future work the other approaches in cluster method viz., centroid method and maximum likelihood method may be attempted to make a comparative study of the extraction of QoS parameters for its effectiveness.

Keywords

Factor Analysis, Principal Components, Quality of Service, Varimax Rotation, Web Service Selection.

Full Text

   

F. Ezhil Mary Arasi ¹, S. Govindarajan ¹, A. Subbarayan ¹

Affiliations
1 Department of Master of Computer Applications, SRM University, Kattankulathur, Chennai - 603203, Tamil Nadu, IN

Abstract

Background/Objectives: The primary objective of the study is to evaluate the performance of Composite Web Service based on the parameters relating to Quality of Service by using statistical methodologies. Methods/Statistical Analysis: By using Web Service Crawler Engine we have collected data on eleven parameters of interest from a more number of Web Services. Among the eleven parameters we have selected six parameters of importance using the sampling of thirty Web Services. We have constructed the weight matrix for the parameters. By using Web Services Relevancy Function ranking has been done for three different scenarios. Using Spearman's Rank Correlation Coefficient we have compared the scenarios. Findings: The graphical representation of the three scenarios shows varying pattern. The Spearman's Rank Correlation coefficient of the Web Services Relevancy Function varied widely among the sampled Web Services. F-Test revealed that there are no significant differences in respect of weights of Quality of Web Services for three different scenarios. We concluded that weight assigned for Quality of Service parameters in Scenario I is preferred than the scenarios II and III through the evaluations. Application/Improvements: The technique developed in this research paper can be applied for the comparison of Web Services during discovery to enhance the performance of the Composite Web Service.

Keywords

F-Test, Quality of Service, Rank Correlation, Selection, Web Services.

Full Text