A new data organizing algorithm for parallel searching

This article focuses on the problem of distributing a data base (i.e., a set of records) in a practical distributed computer network system to facilitate parallel searching. In this distributed data base computer network system, we assume that all records are stored in nodes. Whenever a query occurs, all nodes are searched concurrently. In this article, we introduce a concept widely used by statisticians - the factor analysis technique. We show that the factor analysis technique can be used to propose a new allocation method with which multiattribute data records can be allocated onto several nodes such that the maximum node accessing concurrency can be achieved when responding to partial match queries. A mathematical verification and some experimental results show that our method can indeed be used to improve the multinode data allocation efficiency for concurrent accessing. In addition, the simulation studies comparing the proposed method with Chang's heuristic algorithm in terms of the average response time, in response to all possible partial match queries, show the former to be more effective.