How Contextual Data Influences User Experience with Scholarly Recommender Systems: An Empirical Framework

Abstract

Since the advent of Recommender Systems (RSs), many papers have been published with the majority focusing on creating more accurate algorithms. The more accurate the algorithm is, the better the recommendation is predicted to be for users. Recently, RSs researchers pointed out that the embedding of the recommendation methods in the User Experience (UX) dramatically affects the recommender systems’ value to users. This paper proposes a framework to explore how contexts influence UX with Scholarly recommender systems and identifies relevant contexts to be incorporated in the UX. We first review existing models and theories of UX that are most applicable to RSs and identify gaps in existing work. The framework clarifies how contexts can influence UX with SRSs and enriches our conceptual understanding of how contextual information influences UX of scholarly recommender systems. It can serve as a foundation for further theoretical and empirical investigation. An experiment evaluating the user experience is performed using the quantitative method of Partial Least Squares (PLS) Regression and Structural Equation Modeling (SEM) to examine the developed conceptual framework.