Parameter optimization for information retrieval with genetic algorithm

In most of the modern information retrieval (IR) systems, such as Okapi system, there are a variety of parameters to be turned which are data-dependent and sensitive. Manual parameter setting with fixed experimental values is not always feasible and reliable in practical cases. Furthermore, the supervised learning approaches are not applicable for lacking of relevant information while retrieving. Therefore, an automatic unsupervised parameter learning mechanism is necessary and important. In this paper, a genetic algorithm (GA) based parameter optimization approach is proposed and experimented on Okapi system using large scale data sets of TREC11, TREC10 and TREC9 web track collections. It indicates that our algorithm is effective to adjust system parameters and improve the retrieval performance significantly.