A Survey of Metadata Retrieval through Sentiment Analysis
M. Thanga Subha Devi, R. Suji Pramila · 2022 IEEE International Conference on Signal Processing, Informatics, Communication and Energy Systems (SPICES) · 2022
Due to increasing popularity of videos, video sharing platforms play an important role for millions of content creators to watch videos online. Online content publishing platforms such as YouTube, Facebook and Twitch allows content creators to monetize their posted contents/videos. In which metadata such as title, tag, thumbnail, etc. plays a key role in promoting videos. Most of the digital contents are popular due to the interaction between the content creators and YouTube viewers through their owned channels. The sentiment of each metadata is empirically examined for live streaming, which is spread over 25,000 channels containing across 7 million videos. Metadata which impact the video’s view count are number of subscribers, day one view count, video thumbnail, keywords, Google hits, video category, title’s upper-case letters count and length of the title. These features are mainly used to find the popularity and trendiness of a video. Popularity of a digital content or a video is increased by optimizing metadata after a video is posted. This paper focuses on extracting and analysing the metadata from a video which is then followed by a decision making on the video content. Sentiment analysis is used for such a classification.