Design And Implementation On Joint Retrieval Sub-System For Multi-Data-Source In Wide Area Network

Based on the theory of distributed data retrieval, according to the WAN (Wide Area Network) characteristic of low bandwidth, high delay and weak stability, a multi-data-source sub-system has been designed and implemented for database and whole-length indexing to combine multi-database and whole-length multi-literature from different places with each other. This system has presented an improved B-tree index with the length of data nodes and the methods of inserting, deleting and query of relevant records and parallel indexing of multi-index. With the mechanism of user-conversation and the techniques of data caching, parallel indexing, results combination and efficient page separation of multi-database has been implemented. A combination algorithm of pre-cache of result set based on whole-length indexing in WAN has been designed. The techniques such as whole-length indexing, Chinese word stopping, network crawling, webpage processing and service integration, distributed affair management and data window have been studied. With the mechanism of user conversation and data caching, the technique of parallel retrieval for multi whole-length index from different places has been implemented. And also, functions of data source retrieval service which can be registered and discovered and log influx service with technique of message middleware have been implemented, which forms a complete joint retrieval subsystem for multi data-source and solves the problem of multi data-source in WNA. This system has characteristics of high query efficiency, good scalability, strong fault-tolerance mechanism and being unaffected by the fault of network, which satisfies users'requirement of information retrieval.