The allocation of documents in multiprocessor information retrieval systems: an application of genetic algorithms

The determination of an optimal document allocation within a distributed memory, distributed input/output (I/O) multicomputer, the multiprocessor document allocation problem (MDAP), is an NP-complete problem. Obtaining an optimal document allocation, therefore, is computationally intractable, and, hence, heuristic approaches are required. A genetic-algorithm-based approach to MDAP is described. A proof convergence for the algorithm is provided. Some experimental results from a simulation study of the approach are given. The results illustrate the potential of the genetic-algorithm-based approach as a means of tackling MDAP.<
>