Abstract: A computer-based system recommends food items to a consumer based on food items previously purchased by the consumer and by other consumers. Consumers may purchase food online, such as by purchasing meals from the online menus of restaurants. The system analyzes words in the names and descriptions of the purchased meals to develop profiles for the consumers, where each consumer's profile represents that consumer's food preferences. In particular, if a word (such as “chicken”) or multi-word term (such as “pasta primavera”) appears more frequently in the purchase history of one consumer than in the average purchase history of all consumers tracked by the system, then the system concludes that the consumer prefers the corresponding food item. The system recommends that the consumer purchase food items whose names and/or descriptions contain the same or similar words as the food items that the consumer's profile indicates are preferred by the consumer.