The Influence of the Content Recommendation Algorithm in User Viewing Behavior on the Short Video Platform

Abstract

This study explores the effects of content recommendation algorithms on user viewing behavior on short video platforms, with a particular focus on TikTok and similar platforms. By analyzing large data sets and using a variety of statistical methods, by investigate how different recommendation algorithms affect users' viewing time, interaction patterns, and platform engagement. Research shows that personalized recommendation algorithms significantly affect user behavior, and hybrid recommendation systems show better performance than traditional content-based and collaborative filtering methods. The findings show that entertainment content generally generates higher engagement metrics than educational content and that different content types elicit different user responses. In addition, this paper proposes a new multimodal hybrid recommendation algorithm (MMHRA) that integrates content-based recommendation, collaborative filtering, and multimodal learning techniques. Experimental results show that MMHRA has higher accuracy (92%), recall rate (90%), and F1 value (91%) than traditional recommendation algorithms. This study helps people to understand the role of recommendation algorithms in short video platforms and improve the user experience and content delivery of short video media platforms.