Data modalities, consumer attributes and recommendation performance in the fashion industry

Abstract

This paper investigates determinants of recommendation systems’ performance in an online experiment in a large European Internet footwear store. By combining transactional data and archival customer records, a unique database was compiled from which proxy variables were extracted to represent dimensions of consumer loyalty and shopping involvement. These variables were combined in regression analysis with technical characteristics of two types of algorithms employed for generating recommendations: the EMDE algorithm, relying on the LSH method, and the industry-standard CF-RS. Statistical analysis reveals that recommendations are more successful when visual data modality is combined with behavioural data. Better recommendation performance was found to be associated with lower levels of consumer involvement in shopping, as well as higher levels of trust and engagement with the vendor. Experience with the vendor showed a negative correlation with recommendation performance through both its main effect and by its interactions with other consumer-related variables.

Electronic Markets on platform culture

“…The authors Sylwia Sysko-Romańczuk, Piotr Zaborek, Anna Wróblewska, Jacek Dąbrowski and Sergiy Tkachuk explore how the performance of recommendations (i.e., a recommendation leading to the purchase of a recommended product within the subsequent 24 h), is affected by different combination of three data modalities (i.e., consumer behavior data, product information, visual inputs). In their online experiment among Romanian users of a footwear platform, they outlined that the visual modality was the most influential for predicting shoppers' behavior followed by behavioral data (Sysko-Romańczuk et al, 2022). At the same time, the authors stress that designers of recommender systems need to acknowledge the influence of factors, such as the prior experience with the vendor and the inclination of buyers to assess products themselves, and that more nuanced strategies deem appropriate.…”

Computational Model of Recommender System Intervention

“…In these studies, Fayyaz et al [62] has identifed fve major limitations with RS namely Cold-Start (lack of sufcient data availability within the system would cause the RS to be inefcient), Data Sparsity (user-item matrix information would cause the RSs to provide unstable and unreliable recommendations), Scalability (choice among large pool of recommendations to the user), Diversity (selection of the most preferable among large pool of recommendations to the user) and Habituation Efect (presentation manner and platform to the user infuence efcient of the RS). Additionally, studies such as Badewi et al [63]; Sysko-Romańczuk et al [64]; Li et al [65]; Aljukhadar and Senecal [7] have argued that RS users' usually experience psychological reactance which lead to refusal, ignore and disregard of recommendations. Tis argument is further maintained by Ma et al [6] and Aljukhadar and Senecal [7] that RS recommendation rejection is associated with users' psychological reactance which is the cause of unsuccessful RS interventions.…”