크롬 익스텐션을 활용한 유튜브 필터버블 현상 개선

Abstract

Recently, YouTube usage in Korea has reached 72% of the total SNS usage, with 44% of users utilizing YouTube for news consumption. Furthermore, media outlets are sharing content on YouTube, and politicians are using it as a means of communication with the public. As a result, the influence of YouTube is steadily growing, leading to various academic studies on news consumption through YouTube. Among them, there is growing concern about the filter bubble phenomenon, where YouTube's recommendation algorithm reinforces users' ideological tendencies, resulting in polarization. This paper raises concerns about the filter bubble phenomenon on YouTube in the field of current politics and proposes a practical solution using a Chrome extension. The objective is to reduce confirmation bias by exposing users to diverse political content. This paper trains YouTube accounts with conservative/progressive tendencies and analyzes the political inclinations of recommended videos using text mining to confirm the existence of the filter bubble phenomenon. Subsequently, it verifies whether current affairs/politics videos are recommended in equal proportions on the accounts with the applied Chrome extension. The purpose of this paper is to validate the filter bubble phenomenon caused by YouTube's recommendation algorithm and reduce selective exposure through the Chrome extension. This paper holds significance as an empirical study that demonstrates the utility of resolving the YouTube filter bubble phenomenon.