Optimizing Question-Answering Systems Using Genetic Algorithms
Ulysse Côté Allard, Richard Khoury, Luc Lamontagne, Jonathan Bergeron, François Laviolette, Alexandre Bergeron-Guyard · The Florida AI Research Society · 2015
In this paper, we consider the challenge of optimizing the behaviour of a question-answering system that can adapt its sequence of processing steps to meet the information needs of a user. One problem is that the sheer number of possible processing sequences the system could use makes it impossible to conduct a complete search for the optimal sequence. Instead, we have developed a genetic algorithm to explore the space of possible sequences. Our results show that this approach gives the system the adaptability we desire while still performing better than a human-optimized system.