Order-based Fitness Functions for Genetic Algorithms Applied to Relevance Feedback

Recently there have been appearing new applications of genetic algorithms to information retrieval, most of them specifically to relevance feedback. The evolution of the possible solutions are guided by fitness functions that are designed as measures of the goodness of the solutions. These functions are naturally the key to achieving a reasonable improvement, and which function is chosen most distinguishes one experiment from another. In previous work, we found that, among the functions implemented in the literature, the ones that yield the best results are those that take into account not only when documents are retrieved, but also the order in which they are retrieved. Here, we therefore evaluate the efficacy of a genetic algorithm with various order-based fitness functions for relevance feedback (some of them of our own design), and compare the results with the Ide dec-hi method, one of the best traditional methods.