Minimization of algorithmic bias in recommender systems as a factor in enhancing inclusivity and effectiveness of marketing communications

Анотація

The relevance of the study is determined by the rapid expansion of recommender systems within the digital marketing environment, where algorithmic personalization increasingly shapes consumer interaction with brands, information visibility, and the overall effectiveness of marketing communications. While machine learning–based recommendation algorithms enhance content relevance and marketing performance, they also pose risks of algorithmic bias, manifested in reduced content diversity, the concentration of digital attention, and unequal market access for different actors. The purpose of the article is to substantiate approaches to minimizing algorithmic bias in recommender systems in order to strengthen the inclusiveness of the digital environment and improve the effectiveness of marketing communications. The study applies methods of system analysis, comparative analysis, generalization of scientific approaches to marketing personalization, structural and functional analysis of recommender algorithms, and logical synthesis to identify key factors in the formation of algorithmic bias and mechanisms for its mitigation. Results. The role of recommender systems in transforming digital marketing has been examined, and their decisive influence on shaping consumer behavioral patterns has been established. It has been revealed that optimizing algorithms solely on short-term engagement metrics leads to popularity bias, filter bubbles, and decreased inclusiveness in the digital information space. The necessity of transitioning from narrow personalization models to balanced recommendation approaches combining relevance, diversity, and fair exposure has been substantiated. Conclusions. It has been determined that algorithmic bias represents a systemic characteristic of modern recommender technologies affecting both marketing performance and the competitive structure of digital markets. Prospects for further research include the development of quantitative indicators to measure inclusiveness in recommender systems, the investigation of the long-term behavioral effects of personalization, and the analysis of regulatory and ethical frameworks for artificial intelligence that influence the evolution of digital marketing ecosystems.