Designing recommendation agents as extensions of individual users: similarity and identification in web personalization

Proliferation of personalization technologies online results in an increasingly large number of choices. In addition to the choice of products and services, a consumer is also faced with the choice of the recommendation technology itself. A recommendation agent (RA) is a mediator between a user and a vendor. Existing literature seems to focus on designing RAs as a persuasion tool, which more likely focuses on the vendor’s needs. As an alternative, this paper suggests a theory-grounded model of designing an RA as an extension of a user. This paper examines a new construct, user’s identification with an RA, as an indicator of connectivity in the user-IT relationship. The paper explores different effects of surface similarity (e.g. attribute-based RA which is widely used now) and deep similarity (e.g. value-based RA which rarely exists now) on cognitive and emotional identification with the RA and contributes by exploring theoretical and practical considerations of user-centered design.