Research Of Multi-dimensional Qos Model Supporting Web Service Queries

With the rapid develpoment of Web, there are more and more Web services available for users to choose from. However, it’s difficult for users to find the most qualified service from the candidate Web services because the traditional management models of Web services lack of QoS support, and the existing methods of QoS management have some obvious disadvantages. In respect of functions, it can not determine intuitive query intervals or adjust the query interval dynamically or browse the query results by categories. Its efficiency should be further enhanced, for it is not suitable for querying from a mass of Web services.We propose a new Web service management model--MQM (Multi-dimensional QoS Model), which is used to support service inquiries. First, MQM is a hierarchical model. Concept hierarchy of dimension makes it possible to query more intuitive; Second, MQM supports OLAP operations, such as roll-up, drill-down etc. It can adjust the query interval dynamically and allow users to browse the query results by categories. MQM is an effective solution to the functional limitations of traditional models. The model also provides users with an efficient mechanism for Web services querying. There are two kinds of Web services query methods supported by MQM:"Exact Query" and "Fuzzy Query".They are the high-performance replacements of traditional ones, which meet the different needs of users. Supported by query and OLAP operations, MQM can ensure users retrieve the services they need more accurately and faster by QoS. Meanwhile, multi-dimensional model instantiation is also one of the main contents. Part materialized of date cube is intended to achieve the balance between response time and storage space. Not only it reduces inquires response time, but also takes up not too much storage space. Based on Genetic Algorithm, we propose MQM2algorithm used to instantiate MQM partly. This algorithm looks for the optimal solution having deeper dimension level and wider attribute range in a certain storage space limit conditions. By using the computational results of MQM2, part materialized of MQM improves the efficiency of the multidimensional inquires.Finally the proposed model is proved to be effective and efficient by some experiments. It can be seen that MQM has a greater advantage in query performance and user experience. In terms of query time, MQM-based query is much faster than traditional query. In terms of functions, without losing high recall and precision, MQM provides users with a wealth of functions, thereby enabling users to select Web services more convenient.