Explaining recommender systems fairness and accuracy through the lens of data characteristics
EntidadUAM. Departamento de Ingeniería Informática
Fecha de edición2021-07-01
10.1016/j.ipm.2021.102662Information Processing & Management 58.5 (2021): 102662
Financiado porThis work was supported in part by the Ministerio de Ciencia, Innovación y Universidades (reference: PID2019-108965GB-I00), and in part by H2020 PASSEPARTOUT No. 101016956, Servizi Locali S.P.A. Servizi Locali 2.0, and Ministry of Education, University and Research (references: PON ARS01_00876 Bio-D, PON ARS01_00821 FLET4.0, PON ARS01_00917 OK-INSAID)
Versión del editorhttps://doi.org/10.1016/j.ipm.2021.102662
MateriasExplanatory power; Fairness; Accuracy; Collaborative filtering; Data characteristics; Informática
Derechos© 2021 by the authors
Esta obra está bajo una licencia de Creative Commons Reconocimiento-NoComercial-SinObraDerivada 4.0 Internacional.
The impact of data characteristics on the performance of classical recommender systems has been recently investigated and produced fruitful results about the relationship they have with recommendation accuracy. This work provides a systematic study on the impact of broadly chosen data characteristics (DCs) of recommender systems. This is applied to the accuracy and fairness of several variations of CF recommendation models. We focus on a suite of DCs that capture properties about the structure of the user–item interaction matrix, the rating frequency, item properties, or the distribution of rating values. Experimental validation of the proposed system involved large-scale experiments by performing 23,400 recommendation simulations on three real-world datasets in the movie (ML-100K and ML-1M) and book domains (BookCrossing). The validation results show that the investigated DCs in some cases can have up to 90% of explanatory power – on several variations of classical CF algorithms –, while they can explain – in the best case – about 40% of fairness results (measured according to user gender and age sensitive attributes). Therefore, this work evidences that it is more difficult to explain variations in performance when dealing with fairness dimension than accuracy
Lista de colecciones del ítem
Mostrando ítems relacionados por título, autor, creador y materia.