Research A Semantic Caching For Data Manipulating Mediator System

Being one of the important technologies for improving the performance of data access, caching technology has been applied widly in many defferent fields of computing technology successfully. By analyzing the characteristics of the data that users access, replicability and sharing characteristic of data are utilized by caching technology skillfully that can provide the datas to users by coping and saving them at optimal place. Therefore, the dependence on network bandwidth and facilities of high deferring can be reduced so that the efficiency of using data can be improved. By caching the query results and their semantic description information, the query results are organized by semantic cache technology. Corresponding to traditional cache technologies, it can utilize cache more effectively and alleviate workload of the server and network by using the semantic caching technology. Therefore, to improve the response speed and reduce network communications by using the semantic caching technology is the main researching aim of this project.Regarding to mediator system supporting parallel data manipulating as the environment of application and researching, the significance of the caching technology used in mediator system is elaborated. Moreover, the basic idea of applying the semantic caching to mediator systems is proposed and the key technology of applying the semantic caching to mediator systems is deeply reasearched.Semantic caching organization is the important foundation of semantic caching. The utilization and management of cache can be achieved only after the problem of how to organize the cache is solved. In this thesis, the storage content and format are two main searching aspects of the organization of semantic caching, the basic idea of the mechanism of semantic caching is reseached and analysed, and then an algorithm for query processing in cache is designed and implemented. Because the granularity and query processing method of the semantic caching system proposed in this thesis are different from the traditional strategy, which takes the tuple or page as the granularity, a cache replacement strategy is proposed to support this system. Based on the research of the organization of semantic caching, this semantic replacement strategy, combining with the characteristic of the query processing in semantic caching, is researched and designed from 3 aspects, including precision matching, matching and intersecting matching. Access probability is decided by the access state and the access frequency of the description semantic caching item. Based on analying the character of the traditional cache consistency strategy, the new cache consistency strategy fitting the semantic caching mechanism mentioned is proposed. Based on the implement of the prototype of cache system, a series of verification experiments and performance analysis have been done. It is proved by the experiment result that the query performance of the parallel processing mediator system can be notably improved by using this semantic caching strategy.