The mobile office devices market is currently growing, mainly due to the descending cost of wireless technology as well as the high diversity of functions and features covered. Diversity and proliferation become a hard problem when a person or organization aims at selecting the appropriate device for their particular needs. We propose here a framework for producing device recommendations based on personal or business needs. The framework is articulated through an architecture that includes subsystems for data extraction, recommendation and personalization. All these subsystems operate upon a goal-oriented knowledge base whose presentation is the subject of this paper. Our approach is built upon three independent models: a market model, which contains descriptions of the current devices offered in the marketplace; a domain model, which states the needs of the person or organization; and a mediator model, which describes the types of devices available. We show the actors and processes around these models. Last, we present a prototype that acts as proof-of-concept of the recommender system. © Technomathematics Research Foundation.