Data quality in recommender systems: the impact of completeness of item content data on prediction accuracy of recommender systems

Something’s Missing? A Procedure for Extending Item Content Data Sets in the Context of Recommender Systems

“…As recommender systems are an important category of decision support systems (Power et al 2015), this research is in line with recent works which have revealed a significant impact of data quality dimensions, especially completeness, on data-driven decision support systems (e.g., Feldman et al 2018;Heinrich et al 2019;Woodall et al 2015).…”

“…A relevant and widely used category of decision support systems which assists users facing decision-making problems are recommender systems (Porcel and Herrera-Viedma 2010;Power et al 2015). Based on this and in line with Heinrich et al (2019), we refer to the theoretical model for describing the relationship between data quality and decision support systems, presented in Fig. 1.…”

“…This further emphasizes the key role of data (e.g., item content data) for recommender systems to enable individualized decision support for a large number of users in e-commerce settings (e.g., during shopping experiences on e-commerce websites; cf. Heinrich et al 2019;Kamis et al 2010).…”

“…Mladenić and Grobelnik 2003;Vanaja and Mukherjee 2019), as different data sets (e.g., with more or fewer attributes) may lead to varying results of decision support. Thus, the impact of the data quality of data values on different evaluation criteria of decision support systems such as decision quality or data mining outcome has been studied in existing literature (e.g., Bharati and Chaudhury 2004;Blake and Mangiameli 2011;Feldman et al 2018;Ge 2009;Heinrich et al 2019;Woodall et al 2015). Yet, this research neither focuses on how to systematically achieve more complete item content data sets nor on how to define a wellfounded procedure, but instead tries to explain the relationship between data quality and evaluation criteria of decision support systems.…”

“…Feldman et al (2018) propose an analytical framework to investigate the effects of incomplete data sets on a binary classifier that serves for decision support. Heinrich et al (2019) examine the impact of the amount of available attributes and attribute values on the prediction accuracy of recommender systems.…”