Quality enhanced hybrid youtube video recommendation based on user preference through sentiment analysis on comments – a study on natural remedy videos

With the rapid development of technologies, the Internet has become the primary source of information. In recent years, blogs and videos on the Internet have become the primary source of health-related information. About four out of five people trust the online information posted by anonymous people and access it for their own benefit. More specifically, the simple home remedies or natural remedies for minor health issues available on YouTube gain more attention among all age groups. Numerous viewers not only watch them; they even try or use these remedies personally for magical cures. However, the quality of the information about simple treatments and home remedies available on YouTube may not be accurate, reliable, useful, or trustworthy. Moreover, it is not simple to distinguish reliable information from misleading, false information. Thus, the main objective of this research work is to provide a quality enhanced recommendation of YouTube videos based on the user's preferences. In order to achieve the objective, the proposed work computes the scores for content relevancy and content reliability. The analysis of content reliability is two-fold, with implicit evaluation using structural analysis and explicit evaluation via sentimental analysis on YouTube video comments. The experimental and result analysis performed for the proposed work shows that the predicted ratings and the order of recommendations are more useful and have improved performance on reliability of the video content.