With the rising popularity of online shopping, stores can now offer their full catalog of items through a web presence. As the number of choices increases, potential customers need a way to navigate and discover items that may interest them. Artificial immune systems have been used as the base for recommender system construction. Previous implementations use techniques such as collaborative filtering which suffer from scaling problems. In this thesis we propose the construction of a recommender system where each item is encoded into a bit-string that captures relevant attributes. Combined with the r-contiguous bit matching rule, we are able to recommend items based on past purchases by a customer.