Using machine learning and artificial intelligence to analyze audience media consumption: opportunities and challenges

Abstract

The article explores the prospects of using machine learning and artificial intelligence (AI) methods for multidimensional analysis of audience media consumption patterns. The relevance of the topic is due to the growing fragmentation of the media landscape, the personalization of information diets and the need to move to data-oriented media management models in the digital age. The aim of the work is to identify the key opportunities and limitations of using advanced data analytics to identify preferences and behaviors of media consumers. The methodology includes a systematic review of the literature, a series of expert interviews (N=20), as well as an experiment to develop a prototype of a recommendation news system based on machine learning algorithms. As a result, the main directions of using AI methods in media audience analysis are identified: user segmentation, content personalization, forecasting the popularity of materials, identifying trends and anomalies. The accuracy of recommendation models has been achieved at the level of 84-93%. Problems related to the quality and integration of data, "black boxes" of algorithms, and amplification of echo chamber effects have been identified. A conceptual model of the incorporation of AI tools into the management cycle of a media enterprise is proposed. The need is formulated to develop approaches to assessing the socio-economic effect of algorithmic solutions, studying their impact on the diversity of content and information bubbles of users.