Current Trends on Sentimental Analysis on Youtube Videos

YouTube, one of the foremost watched video sharing websites with millions of different types of contents and their creators with innovative ideas was originated in USA. You Tube has been recorded as the second largest program watched, searched and time spending platform by the millions and billions of views after Google. The videos uploaded by the creators on You Tube majorly depend on the content, knowledge and skilled dynamics and social interaction between the channels (content creators) and user of it. Social networks like videos spreads and sensitivity of YouTube meta-level features provide an important impact on the popularity of videos. Emotions are indeed for social interactions. To build automatic speech recognization system, emotions play a vital role. poor data quality, public dataset, manual labelling, CPU bottlenecks, model over fitting, exploding gradient problem and Data storage are major challenges in the sentimental analysis from the videos. This paper describes the various approaches for sentiment analysis of videos and about the algorithms which provide quite accurate results and to improve the performance rate, parallel processing can be done in Graphical Processing Unit (GPU) using NVIDIA Compute Unified Device Architecture (CUDA) language.