E-commerce firms are increasingly adopting recommendation systems to effectively target customers with products and services. The first essay examines the impact that improving a recommender system has on firms that deploy such systems. A market with customers heterogeneous in their search costs is considered. We find that in a monopoly, a firm always increases its profit by improving its recommender system; interestingly, decreasing the price may be optimal when a small reduction in price significantly increases the market size of the firm. In a duopoly with firms asymmetric in their recommender system effectiveness, we show that both firms may find it optimal to "reduce" their prices simultaneously under certain conditions. When the firm with recommender system of higher effectiveness improves its system, both firms increase their profits by increasing their prices. In the second essay, we develop an algorithm to generate a novel kind of association rules, referred to as disjunctive consequent rules that are suitable for recommending multiple items for web-based retailing. Traditional rules with single or conjunctive consequents are of limited use in such environments. Experiments conducted rules on real datasets show that the accuracies of recommendations made using disjunctive consequent rules are significantly higher than those made using traditional association rules. The third essay develops a probabilistic framework for combining association rules in order to base recommendations on as many items of a customer's basket as possible. Traditional rule-based systems typically rely on identifying one among several eligible rules in order to make recommendations, ignoring information from other eligible rules that can potentially improve recommendations. We present a probability-based approach to combine multiple rules when making recommendations. When multiple combinations exist, we propose a maximum-likelihood approach to identify the best combination. Experiments show that the accuracies of recommendations improve significantly when items are recommended using multiple rules instead of single rules. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]