Evaluation of Facebook advertising recommendations explanations with the perspective of semiotic engineering

Abstract

Recommendation systems (SRs) have a major impact on what people consume daily. In order to achieve the goal of recommending relevant items, in addition to improving the accuracy of the recommendation algorithms, it is necessary to use user-centered approaches, so that SRs can inspire trust and loyalty, increasing users satisfaction, and convincing them to recommended item. One way to achieve these objectives is to explain the recommendations generated by these systems. An explanation starts from a communicative process between a sender and a receiver. Thus, in addition to user-centered approaches, it is also necessary to think of a communication-centered approach. In this work, we propose to explore and reflect, from a Semiotic Engineering perspective, on how the recommendation systems explanation process happens. Thus, we carried out an evaluation of the quality of the explanation given by Facebook regarding the advertising recommendations offered in the users timeline, based on the Semiotic Inspection Method and the application of a questionnaire.