Loading [a11y]/accessibility-menu.js
Popularity Inference Based on Semantic Sentiment Analysis of YouTube Video Comments | IEEE Conference Publication | IEEE Xplore

Popularity Inference Based on Semantic Sentiment Analysis of YouTube Video Comments


Abstract:

YouTube is a popular online video-sharing and social media platform, hosting diverse content for all ages. The platform is gradually gaining ground against traditional me...Show More

Abstract:

YouTube is a popular online video-sharing and social media platform, hosting diverse content for all ages. The platform is gradually gaining ground against traditional media, with major turnover increasingly turning there due to videos and creators' influence, especially in younger audiences. Some of the videos may attain a large number of views often accompanied by many comments. Comments, often focus not only on the content of the video but also on several other aspects and topics. In this paper, we focus on the sentiment analysis of video comments, using a variety of methods we explore the correlation between user comments' sentiment polarity score and the popularity of the relevant video. Some adversities of machine learning or neural network model approach to the task include linguistic irregularities, typing errors, niche vocabulary, sarcastic and ambiguous content, or unrelated events influencing the commented themes. Unlike previous works [1], [2], our focus is centered on leveraging comment sentiment polarity for deriving predictions of the popularity and success of the respective videos. Our methods depend on lexicon-based sentiment analysis approaches like VADER and TextBlob but also on neural network architectures including BERT and RoBERTa. Our experimental analysis resulted in promising results, setting a baseline for this application theme.
Date of Conference: 10-12 July 2023
Date Added to IEEE Xplore: 15 December 2023
ISBN Information:
Conference Location: Volos, Greece

References

References is not available for this document.