Abstract
Electronic commerce (e-commerce) has grown rapidly over the past years. Products, services and information of different types are offered daily for many Internet users. Finding out an appropriate strategy to offer a product to each customer in a personalized fashion is the goal of a recommender system. This association between items is a task that falls under the umbrella of data mining, more specifically the area of association rule mining, or simply, association rules. This paper investigates the use of evolutionary algorithms and artificial immune systems to build associations of items in real-world e-commerce databases. The performances of the evolutionary and immune algorithms for rule mining are compared to each other. A discussion is made in terms of a group of interest measures of associations and computational time necessary to learn the rules.