Automated-Sentiment-Analysis-of-Web-Multimedia

The popularization of multimedia content on the Web has raised the need to analyze and retrieve it automatically. Manually labeling these data is extremely expensive and unfeasible; therefore, automatic methods for large-scale identification of sentiments are needed. For example, podcast search and retrieval benefits from the use of sentiments related to the content, but most of the multimedia being shared are published without any relevant information to identify them. Moreover, due to this, users are poorly aware of the contents of multimedia. This results in people uploading content solely for monetization purposes and click-baits. Hence for this purpose, we decided to automate the content identification using Audio Processing and Machine Learning.